Skip to main content

Full text of "Consumer satisfaction: an extended research conceptualization"

See other formats


CONSUMER SATISFACTION: 
AN EXTENDED RESEARCH CONCEPTUALIZATION 



DOUGLAS R. HAUSKNECHT 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 

OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 

OF THE REQUIREMENTS FOR THE DEGREE OF 

DOCTOR OF PHILOSOPHY 

UNIVERSITY OF FLORIDA 

1988 






This dissertation is dedicated to the memory of my friend, Michael 
Ray Albaugh. Mike was among my first students; one who made teaching 
worthwhile. He displayed a keen interest and ready aptitude for 
marketing. These traits were demonstrated further when he assisted in 
the collection of the main study data, reported in these pages, as part 
of an independent study course. I deeply regret the fact that this 
document was not completed at the time of his death in an Akron, Ohio, 
rooming house fire May 14, 1987. Mike was graduated from the 
University of Akron as Bachelor of Science in Business Administration/ 
Marketing on May 31, 1987. He had completed all of the requirements 
for the degree. 



ACKNOWLEDGMENTS 

Any undertaking of this magnitude is completed only with the 
assistance and cooperation of others. So many people have been so 
helpful over such a long period, I hesitate to identify any in fear of 
omitting others. 

The most direct influence on a dissertation is, of course, the 
committee chairman. Rich Lutz provided not only guidance and 
direction, but also the motivation and encouragement that were 
necessary to keep things moving. Most importantly, he helped me to 
develop the confidence and perception of self-worth that enabled me to 
achieve closure at last. 

Joe Alba and Sam Ahmed, the other official committee members, kept 
me from going astray in designing and conducting the research. For 
this, and their willingness to adapt to tight schedules and too-often- 
missed deadlines, I am grateful. 

In a very real sense, each member of the marketing faculty has 
contributed to the dissertation over the years. Whether as the result 
of in-class instruction, seminar presentation, or informal discussion, 
each has influenced the research compiled herein. I am truly the 
product of the entire faculty. 

Some deserve special recognition. Joel Cohen planted the seed 
that caused me to consider doctoral studies. Even before that, he set 
education and career matters in their proper perspective by helping me 



learn an unforgettable lesson about life's priorities. Dipankar 
Chakravarti and Chezy Ofir introduced me to the details of research and 
the academic life. Without this, I would likely never have pursued 
this career. Bill Wilkie first acquainted me with and encouraged my 
perspectives on postpurchase topics in consumer behavior. 

I should also acknowledge the contributions of academics with whom 
I had no formal connection, but who were willing to provide the 
feedback and advice that I needed. Many fall into this category, but 
Rich Oliver stands out for the effort expended on my behalf. 

Professional debts are only part of the story. When a 
dissertation takes over life, it affects and is affected by all of the 
individuals who have provided encouragement along the way and have 
responded with patience when that was most needed. My family, 
immediate and extended, bore the brunt of this burden. In addition, 
Jim and Sunny Flavin are the friends who started me on this path and 
have walked with me along the way. Ron and Judy Albaugh provided 
incentive for the final stretch. 

Finally, understanding, patience, and assistance were supplied at 
the University of Akron — where I began teaching midway through the 
research. Faculty colleagues, especially Judy Wilkinson, have given 
advice and encouragement. Staff assistance, especially the typing of 
Pat Johnson, made the feat manageable. Both the University of Akron 
and the University of Florida provided partial financial support of the 
research. 

There are others who have contributed much. They may be left off 
the printed page, but they are etched indelibly in memory. 



TABLE OF CONTENTS 



page 



ACKNOWLEDGMENTS iii 

LIST OF TABLES vii 

LIST OF FIGURES ix 

ABSTRACT x 

CHAPTERS 

1 INTRODUCTION 1 

Purpose 1 

Importance of Research in Consumer 

Satisfaction 2 

Audiences Addressed by the Research 4 

CS/D Literature to Date 6 

Contribution of Dissertation 7 

Dissertation Overview 8 

2 LITERATURE REVIEW 10 

Introduction 10 

Definitions of Satisfaction 11 

The Disconf irmation Paradigm 21 

Other Perspectives on Satisfaction 33 

Types of Consumer Satisfaction 33 

Attitudinal-Behavioral Consistency 44 

Measurement Issues in the CS/D Literature 47 

Behavioral Outcomes of Satisfaction 64 

Summary 78 

3 SATISFACTION: AN EXTENDED CONCEPTUALIZATION OF 

THE GENERAL CONSTRUCT 80 

Introduction 80 

Definition and Theory of Emotions 81 

Motivation 83 

Lack of a Satisfaction Feeling 91 

Other Types of Satisfaction 95 

The Relationship Between Satisfaction and 

Attitude 98 

The Extended Conceptualization 102 

Purpose of the Extended Conceptualization 142 

Summary and Conclusions 165 



4 RESEARCH HYPOTHESES 

Introduction 

Main Hypotheses 

Secondary Hypotheses 

Supplemental Hypotheses 

Summary 

5 EXPERIMENTAL METHOD 

Introduction 

Pilot Study--Measure Development 

Research Conducted 

Summary 

6 STUDY RESULTS 

Introduction 

Pretest Findings 

Main Study Results 

Summary and Discussion 

7 CONCLUSION 

Introduction 

Summary of Findings 

Implications of Findings 

Future Outlook 

APPENDIX A PILOT STUDY QUESTIONNAIRE 

APPENDIX B INSTRUCTIONS FOR ADMINISTRATION OF EACH 
EXPERIMENTAL CONDITION 

APPENDIX C QUESTIONNAIRE AND SUBJECT FORMS FROM 

MAIN STUDY 

REFERENCES 

BIOGRAPHICAL SKETCH 



167 

167 
167 
169 
170 
172 

174 

174 
174 
195 
234 

235 

235 
235 

241 
339 

342 

342 
342 
351 
354 

357 



365 

397 

413 

452 



LIST OF TABLES 

Table Page 

1 . Perspectives on Consumer Satisfaction 29 

2. Types of Consumer Satisfaction 40 

3 . Actions Following Dissatisfaction 76 

4 . Measures of Fundamental Emotions 85 

5. Patterns of Emotions in Automobile Ownership 86 

6 . "Triggers" of Response Involvement 132 

7. Prior CS/D Research and the Extended Conceptualization... 146 

8. Results of Factor Analysis of Emotion Items 180 

9. Behavioral Intent Scales from Factor Analysis 183 

10. Reliability Coefficients for Scales Used in 

Pilot Study 186 

11. Adjectives Describing Ends of Satisfaction Continuum 137 

12. Mean Responses by Condition and Classification 190 

13. Mean Responses to Likert Items by Condition 

and Classification 192 

1 4. Manipulation Summary 199 

15. Preparation of Peanuts for Coffee Blend 212 

16. Coffee Preparation Instructions 220 

17. Measure Summary 223 

18. ANOVA Summary of Decision 243 

19. ANOVA Summary of Coffeemaker Decision 244 

20. Selection of Coffee Discrepancy Scale 245 

21. Selection of Coffeemaker Combined Discrepancy Scale 246 

22. ANOVA Summary of Coffee Discrepancy 248 

23. ANOVA Summary of Coffeemaker Combined Discrepancy 249 

24. ANOVA Summary of Coffeemaker Discrepancy 250 

25. ANOVA Summary of Coffeemaker Expectancy Disconf irmation. . 251 

26. Selection of Coffee Expectations Scale 252 

27. ANOVA Summary of Coffee Bitterness Expectations 253 

28. Selection of Coffeemaker Expectations Scale 255 

29. ANOVA Summary of Composite Coffee Expectations 257 

30. ANOVA Summary of Coffeemaker Expectations 258 

31 . Selection of Coffee Perception Scales 260 

32. Correlations Among Coffeemaker Perception Scales 262 

33. ANOVA Summary of Coffee Quality Perception 263 

34. ANOVA Summary of Coffee Color Perception 264 

35. ANOVA Summary of Coffee Flavor Perception 267 

36. ANOVA Summary of Coffee Bitterness Scale 263 

37. ANOVA Summary of Coffee Flavor (single) Scale 269 

38. ANOVA Summary of Coffeemaker Speed Perception 271 

39. ANOVA Summary of Coffeemaker Filling Perception 272 

40. ANOVA Summary of Coffeemaker Use Perception 273 

41 . ANOVA Summary of Coffeemaker Appearance Perception 274 



List of Tables — continued 

[•able Page 

42. Selection of Coffee Involvement Scale Combinations 277 

43. Selection of Coffeemaker Involvement Scale 

Combinations 278 

44. Results of Response Involvement Analyses 280 

45. Reliability of Coffee Attitude Scale 283 

46. ANOVA Summary of Coffee Attitude Ratings 284 

47. Reliability of Coffeemaker Attitude Scale 288 

48. ANOVA Summary of Coffeemaker Attitude Ratings 289 

49. Correlations Between Perception and Attitude Measures.... 291 

50. Significance Tests for Perceived Discrepancy Measures.... 294 

51. ANOVA Summary of Coffee Satisfaction Ratings 302 

52 . ANOVA Summary of Coffee Odds Scale 303 

53. ANOVA Summary of Coffee Delighted-Terrible Scale 304 

54. ANOVA Summary of Coffee Faces Scale 305 

55 . ANOVA Summary of Oliver ' s Six Item Scale 305 

56. Reliability of Oliver's Six Item Scale 309 

57. ANOVA Summary of Coffeemaker Satisfaction Ratings 310 

58. ANOVA Summary of Coffeemaker Odds Scale 311 

59. ANOVA Summary of Coffeemaker Delighted-Terrible Scale.... 313 

60. ANOVA Summary of Coffeemaker Faces Scale 314 

61 . Number of Subjects Contacted by Telephone at 

One Week Delay 315 

62. Significant Linear Correlations Between Measures of 
Satisfaction, Dissatisfaction and Attitude with 
Hypothesized Behavioral Outcomes 318 

63. Significant Linear Correlations Between Measures of 
Satisfaction, Dissatisfaction and Attitude with 
Hypothesized Behavioral Outcomes Positive 

Discrepancy Conditions Only 319 

64. Significant Linear Correlations Between Measures of 
Satisfaction, Dissatisfaction and Attitude with 
Hypothesized Behavioral Outcomes Negative 

Discrepancy Conditions Only 320 

65. Correlations Between Involvement and Emotion Measures.... 323 

66. Number of Subjects Reaching Criterion Emotion Levels 326 

67. ANOVA Summary of Satisfaction Emotions 327 

63. ANOVA Summary of Dissatisfaction Emotions 328 

69. Reliability of Emotion Measures 330 

70. ANOVA Summary of Interest 333 

71 . ANOVA Summary of Joy 334 

72. ANOVA Summary of Surprise 335 

73. ANOVA Summary of Anger 335 

74. ANOVA Summary of Disgust 337 



LIST OF FIGURES 

Figure Page 

1 . Cardozo's Model of Satisfaction 23 

2. Anderson and Hair Treatment of Satisfaction 24 

3. Model of the Satisfaction Process 27 

4. Measures Used in Consumer Satisfaction Research 49 

5 . Post Evaluation Responses 66 

6. Satisfaction Responses 67 

7. Satisfaction As Emotional Continua 90 

8. Relative Influence of Satisfaction and 

Attitude on Certain Consumer Behaviors 100 

9. An Extended Conceptualization of the General 
Satisfaction Construct 103 

10. Descriptive Stimuli 203 

1 1 . Hypothesized Relationships 230 

12. Perceptions of Coffee Strength Interaction 

Among Experimental Conditions 266 

13. Attitude Toward Coffee Responses by 

Experimental Condition 286 

14. Mean Responses to Standard Satisfaction Scales 296 



Abstract of Dissertation Presented to the Graduate School 

of the University of Florida in Partial Fulfillment of the 

Requirements for the Degree of Doctor of Philosophy 

CONSUMER SATISFACTION: 
AN EXTENDED RESEARCH CONCEPTUALIZATION 

By 

DOUGLAS R. HAUSKNECHT 

December 1988 

Chairman: Richard J. Lutz 
Major Department: Marketing 

Consumer satisfaction/dissatisfaction (CS/D) is conceptualized as 
the antecedent of specific post-consumption behaviors. The 
conceptualization presents a general construct, satisfaction, as the 
feeling that results from six necessary and sufficient conditions: 
expectations , decision, personal experience , perception , response 
involvement , and conf irmat ion/ dis confirmation of the expectations and 
decision concerning the object. 

A measurement technique designed to tap the emotional nature of 
the satisfaction construct is derived from the Differential Emotions 
Scale (DES) using a critical incident methodology. A satisfaction 
feeling is defined as the emotional pattern of interest, joy and 
surprise; dissatisfaction is defined as anger, disgust and surprise. 

An experiment was conducted using 203 subjects from convenience 
intercepts randomly assigned to the cells of a 2x2x2 design. Factorial 
manipulations of positive/negative discrepancy between performance and 



expectations, presence/absence of opportunity to make a decision, and 
low/high response involvement with the test product were 
operationalized using ground roast coffee. Dependent variables 
included the DES, other commonly used satisfaction scales, attitudes, 
behavioral intent and actual behavior. 

Attitude and typical satisfaction scores and behavioral measures 
were consistent with the discrepancy manipulation; the better coffee 
averaged better scores than the poorer quality coffee. These measures 
also tended to indicate an effect of the decision manipulation, with 
free choice making subjects more critical. 

A low level of response involvement was evidenced in the lack of 
effects on the dependent measures and contributed to weakness in the 
measures of emotions. The DES measures of CS/D showed none of the 
predicted patterns. Correlation analyses between emotion measures and 
behavior or behavioral intent measures also displayed little 
consistency. 

The importance of a decision in CS/D studies was demonstrated by 
its influence on most measures. Among the measures attempted, the 
behavioral self-reports indicated promise for future investigation. 
Measures of perceptions versus expectations at the attribute level 
indicated that even a single, brief consumption experience may lead to 
a more complex cognitive structure. 



CHAPTER 1 

INTRODUCTION 

Purpose 

The study of consumer satisfaction/dissatisfaction (CS/D) has been 
recently criticized for the lack of shared constructs (Day 1983, 1980b; 
Swan 1983; Woodruff, Cadotte and Jenkins 1983b) and insufficient 
attention to experimental realism (LaTour and Peat 1979; Russo 1979). 
These problems lead to difficulty in making comparisons across studies 
or even in drawing conclusions from a single study. 

The dissertation addresses these concerns in several ways. The 
general construct of "satisfaction" is defined to be a feeling, 
distinct from attitude, with separate and distinguishable influences on 
behavior. The application of this construct to marketing situations 
defines the area of interest known as consumer satisfaction. Within 
this field of study, the object of the feeling may be a particular 
product/use occasion, 1 the brand in general, or the product class in 
general (that is, the object can be defined at any level of 
specificity — as long as it is defined consistently across the 
conceptualization and measurement of the study) . 

Satisfaction feelings do not, however, invariably result from all 
consumption situations. Rather, there must be certain key factors 



1 The word "product" is used generically throughout the 
dissertation and will generally encompass services and other consumer 
goods . . 

1 



which serve to generate, or at least activate, the mechanisms that 
arouse the emotions. When one or more of these factors is missing, the 
measurement of satisfaction as a distinct feeling should automatically 
be suspect. The dissertation presents these situational factors as 
necessary and collectively sufficient antecedent conditions for the 
arousal of the satisfaction feeling. The primary contribution of the 
conceptualization is the delineation of a theoretical network which 
integrates (1) the construct described by Oliver (1981); (2) the 
model proposed by Day (1983); (3) a measurement approach (Westbrook 
1983) that operationalizes the construct as defined; and (4) the 
influence on the consumer behaviors that have been determined to be cf 
interest. One useful result of this theorizing and research is a 
template against which to compare empirical work on ''consumer 
satisfaction" to see whether the construct is justifiably invoked. 
Finally, the satisfaction research in consumer behavior is examined 
vis-a-vis satisfaction research in other disciplines and a case is 
built for a general construct and a general theory of satisfaction in 
which the field of consumer satisfaction is positioned and to which the 
definition and framework presented apply. 

Importance of Research in Consumer Satisfaction 
"Customer satisfaction with a product presumably leads to repeat 
purchases, acceptance of other products in the same product line, and 
favorable word-of-mouth publicity. If this assumption is correct, then 
knowledge about factors affecting customer satisfaction is essential to 
marketers" (Cardozo 1965, p. 244). This is the premise with which one 
of the seminal papers in consumer satisfaction began. That is, there 



is some construct , referred to as satisfaction, that has unique 
effects on behaviors that are of great interest to marketers. In fact, 
Marketing has been defined as ". . . the profitable creation of 
customer satisfactions" (Risley 1972, p. 10). This investigation falls 
within a growing literature that deals explicitly with postpurchase, 
consumption behavior (Belk 1984b). 

The Marketing Science Institute's 1976 conference on Consumer 
Satisfaction/Dissatisfaction (CS/D) was, in part, an attempt to address 
the need for rigorous conceptualization and definition (Hunt 1977a). 
Yet, such features are still lacking in the scholarly research in the 
area (Day 1983). If satisfaction is to have any value as a separate 
construct, then its definition and measurement must be distinct from 
other related constructs such as brand attitude, and it should be shown 
to have separable influences on phenomena of interest (MacCorquodale 
and Meehl 1948) . 

The dissertation focuses on satisfaction as a unique construct and 
bases its conceptualization and measurement on this philosophy. The 
equating of satisfaction and attitude (e.g., LaTour and Peat 1979) and 
the operationalism approach (e.g., Aiello and Czepiel 1979) are 
rejected in favor of a more integrative framework in which satisfaction 
intervenes between its antecedents and the behaviors it is presumed to 
affect (Day 1983). Day's framework, however, implicitly assumes that 
the feeling of satisfaction — at some level— always results from a 
consumption situation. The dissertation argues that other antecedent 
conditions must be present before evaluations are made and feelings 
result. Accurate measurement of satisfaction, which includes the 



ability to detect its absence, is therefore crucial to the 
conceptualization being proposed. 

Toward this end, a measurement technique is developed based on the 
emotional measures used by Westbrook (1983; 1987) and is used to 
predict behaviors such as those originally proposed by Cardozo (1965). 
This measurement scheme is (a) bound to the definition and 
conceptualization of satisfaction that are developed herein and (b) 
less susceptible to the experimental demand effects raised by more 
conventional techniques. 

The definition of satisfaction that is adhered to throughout the 
dissertation is the one proposed by Day: "an emotional response 
manifested in feelings and is conceptually distinct from cognitive 
responses, brand affect and behavioral responses. In particular, it is 
not a 'kind of attitude' as suggested by some authors" (1983, p. 113). 

The conceptualization that is proposed elaborates on this 
emotional response as resulting from the presence of certain necessary 
and collectively sufficient antecedents: (1) expectations, (2) 
decision, (3) personal experience, (4) perception, (5) response 
involvement, and (6) conf irmation/disconf irmation. Each of these 
antecedents is defined and discussed in greater detail in the context 
of deriving a general conceptualization that is derived from and 
applies to a range of human behaviors (Belk 1984c). 

Audiences Addressed by the Research 

Many businesses rely to a large extent on repeat purchasers and 
positive word-of-mouth among customers for their continued success. 
Unfortunately, a very common method of assessing (and thereby making 



decisions about) satisfaction has been to monitor complaints and/or the 
absence thereof (Bohl 1987). 

Interest in consumer satisfaction by the federal government also 
has focussed primarily on complaint behavior (Technical Assistance 
Research Programs — hereafter TARP — 1979) and methods to improve the 
effectiveness of consumer complaining (Knauer 1983). This is not too 
surprising inasmuch as consumers who feel they have not gotten an 
adequate response from the company or industry involved can and 
sometimes do turn to state or federal agencies and legislators (Smith 
and Bloom 1984) . 

But as will be shown later, complaining — especially to a third 
party after prior attempts at redress have failed — is a relatively 
extreme response that depends on more than just the level of 
dissatisfaction (Richins 1932). Moreover, the absence of 
dissatisfaction sufficient to prompt a complaint is not evidence for 
the presence of satisfaction sufficient to encourage word-of-mouth or 
repurchase. 

Complaining behavior, although not an adequate surrogate for 
dissatisfaction (or satisfaction) measurement, does offer the advantage 
of visibility. The abundance of complaints helps to maintain interest 
in consumerism among academics (Bloom 1982a; Bloom and Greyser 1981), 
authors of popular "How to Complain" books (Charell 1973; Eisenberger 
1977; Horowitz 1979) and members of grassroots consumer groups (Barnes 
and Kelloway 1980) . 

It should also be noted that dissatisfaction and the resultant 
negative behaviors are not the sole focus. Equivalent positive 



behaviors (brand loyalty, complimenting, etc.) are also of interest. 
The seemingly more urgent complaint behavior, however, has garnered the 
bulk of the attention directed toward post-satisfaction behavior. 

The research reported in these pages should be of interest to 
marketing practitioners, regulators, consumers, and consumer advisors. 
Moreover, a consistent research paradigm can clarify the 
interpretability of satisfaction reports by third-party research 
suppliers (Serafin 1987). 

CS/D Literature to Date 

The last decade or so has witnessed a great deal of interest and 
scholarly research activity focused on consumer satisfaction, 
dissatisfaction and complaining behavior (Hunt 1982; 1983). The 
output resulting from this effort ranges across a spectrum of topics 
and details. Yet, despite the sheer volume (Hunt lists nearly 600 
relevant references), there is only qualified agreement on the most 
basic of issues. No real consensus is approached concerning the 
definition of satisfaction, its measurement, or the psychological 
states or processes involved (Day 1983; Swan 1983; Woodruff, Cadotte 
and Jenkins 1 983a) . 

The major consistencies within the field have been (1) the 
treatment of dissatisfaction—satisfaction as a single continuum and 
(2) the adoption of the disconf irmation of expectations paradigm. A 
few authors have proposed two separate satisfaction and dissatisfaction 
constructs, arguing that the attributes that determine each are 
different (Swan and Combs 1976). Research that tests this proposition 
has generated, at best, mixed support in limited circumstances. 



Similarly, alternatives and modifications to the consensus paradigm 
(i.e., that satisfaction is the result of perceived reality exceeding 
or matching expectations) have been discussed in limited contexts. As 
noted by Day (1983), disconf irmation-of-expectations models have been 
applied and tested at both attribute-specific and summary-evaluation 
levels. 

The literature has not addressed government concerns of how to 
increase satisfaction or decrease dissatisfaction (Leavitt 1977), 
industry questions regarding the seemingly conflicting needs to raise 
expectations to generate sales but lower expectations to generate 
satisfaction (Lutz 1980), or even how the research might aid individual 
consumers (Haines 1979). These broader questions can not be adequately 
answered until a broader framework for satisfaction is understood. 
Contribution of Dissertation 

Although we do not yet know enough about satisfaction to propose a 
general theory that can specify the influence of a multitude of 
variables across a range of situations (Russo 1979), the field is well- 
served by attempts such as that by Day (1983) to circumscribe what is 
known within a general framework. The dissertation continues this 
effort by applying his definition and an extended version of the 
framework to the derivation of a prototypical experimental protocol. 
Such a protocol must include each of the necessary antecedent 
conditions so that the satisfaction construct may truly be tapped. 
This represents one step in a movement toward realism and viewing 



consumers as subjects who instigate behavior rather than as objects 
affected by the marketer or experimenter (Olander 1980). 

Further, satisfaction is defined in emotional terms, requiring a 
measurement methodology consistent with such a definition. A set of 
emotions that comprise satisfaction is measured and its relationships 
to behaviors that have been theorized to result from satisfaction are 
explored. Finally, these measures are used to test the necessity of 
three of the antecedent conditions by testing for the presence or 
absence of the crucial emotions when the presence of the antecedents is 
manipulated. 

Dissertation Overview 

Chapter 2 reviews the current state of the CS/D literature. The 
chapter begins with a definition of the satisfaction construct and 
discussion of the paradigm that dominates the research in the area. 
Other approaches to the topic are then discussed, followed by a review 
of the various types of satisfaction that have been considered under 
the guise of consumer satisfaction. The importance of recognition of 
the interrelationships of types of satisfaction is emphasized in 
discussions of measurement issues and behavioral outcomes of 
satisfaction that conclude the chapter. 

In Chapter 3 an extended conceptualization of the construct is 
presented. General discussions of the nature of emotions and 
motivation introduce the chapter, followed by the description of 
satisfaction as an emotional construct not limited to the domain of 
consumer behavior. This latter point is highlighted by a discussion of 
the treatment of satisfaction in other literatures and the related 



concept of attitude. Each of the necessary antecedents is then defined 
and its role in the overall process discussed. Finally, prior research 
in CS/D is evaluated with respect to conformity with the present 
conceptualization and the likely impact on interpretability of the 
findings. 

Chapter 4 presents explicit hypotheses based on the arguments 
previously detailed. The major hypotheses discussed represent tests of 
the necessity of the two of three antecedents that are introduced by 
the dissertation. 

A measurement technique capable of assessing the emotional nature 
of satisfaction is derived in the first half of Chapter 5. Following 
this, a research design is presented that uses emotion and other 
measures to test the hypotheses. 

Chapter 6 details the results of the study. 

Chapter 7 discusses the implications of the study for the new 
conceptualization and outlines future research that builds upon the 
findings . 



CHAPTER 2 

LITERATURE REVIEW 

Introduction 

The topic of consumer satisfaction/dissatisfaction (CS/D) has been 
the subject of much study over the past decade. In this chapter, the 
literature resulting from this attention is reviewed. 

Although the term "satisfaction" has connotative meaning to most 
people, this is not adequate for examining and establishing the 
existence of a theoretical construct. This chapter, therefore, begin? 
with a look at the various definitions of satisfaction that have been 
proffered in the CS/D literature. Next, the disconf irmation 
paradigm — generally considered to dominate CS/D research--is outlined 
and its development is discussed. Other viewpoints on satisfaction are 
also mentioned briefly. 

Because the dissertation treats satisfaction as a general feeling 
that may arise from or focus on a variety of concepts or objects, it is 
necessary to show the diversity of objects that may evoke this feeling. 
These objects of satisfaction are arrayed by level of specificity 
(e.g., particular item versus product class), and implications for 
measurement and theory are discussed. 

One serious shortcoming in the CS/D area that has limited its 
ability to generate a more integrated and better received literature 
has been the lack of an agreed upon measurement methodology. The 

10 



measures that have been brought to bear are examined and their failings 
with respect to satisfaction theory and experimental demand artifacts 
are discussed. 

Finally, CS/D researchers are also implicitly or explicitly 
interested in behaviors that are thought to result from satisfaction. 
The chapter concludes by examining the link between the satisfaction 
feeling and these behaviors (or behavioral intentions). The large body 
of research that examines the consumer complaining phenomenon is 
considered, however, to be a separate literature and is discussed only 
as necessary. 

Definitions of Satisfaction 

Researchers in CS/D have been haunted since the early days of the 

area by the lack of a consensus definition of satisfaction: 

As a concept, at the level of thought, the 
understanding of satisfaction becomes weaker. In a 
simple sense, it lacks substantive meaning — that 
immediate sense of being able to transfer to 
another exactly what it is that is known and 
experienced, how it can be experienced, why it is 
important, and how it is a thought worth sharing 
and doing something about. . . As a concept, 
dissatisfaction has substantive meaning far 
exceeding satisfaction's. Dissatisfaction implies 
many actions: angry consumers, worried managers, 
protesting consumer advocates, and rule-making 
government officials. (Czepiel and Rosenberg 1977, 
p. 93) 

This early uncertainty did not dissipate. During the first CS/D 

conference, satisfaction was defined as (a) the happiness resulting 

from consumption, (b) a cognitive state resulting from a process of 

evaluation relative to previously established standards, (c) a 

subjective evaluation of experiences and outcomes associated with 

acquiring and consuming a product relative to a set of subjectively 



12 



determined expectations, (d) a two-factor process of evaluating a set 
of satisfiers and a set of dissatisf iers, and (e) one step in a process 
including prior attitudes, experience, post attitude and future 
purchase (Day 1980a, Hunt 1977a). Even more recent overviews and 
commentaries on the CS/D literature cite a persistent confusion 
concerning possible definitions (Woodruff, Cadotte and Jenkins 1983b) 
and attempt to provide consensual definitions that would be useful to 
the majority of researchers in the area (Day 1983, Swan 1983). 

As mentioned in Chapter 1, Day's definition has been adopted for 
use within the dissertation. Before elaborating the reasons why that 
definition was chosen, it seems appropriate to discuss the types of 
definition that were eliminated from consideration. Many authors seem 
to eschew theory and proceed directly to operationalizing and testing 
specific hypotheses of interest. This approach evades the question of 
definition and how the definition may fit a larger theoretical 
framework. The CS/D literature is rife with papers that view 
satisfaction through such an operationism perspective. The authors 
implicitly accept the operationalization of satisfaction and its 
measures as the definition of the construct and provide no additional 
conceptualization. Aiello and Czepiel explicitly define satisfaction 
as, "the responses given to simple 'How satisfied are you with . . .' 
questionnaire items with five response categories ranging from 
'Completely satisfied' to 'Not satisfied'" (1979, p. 129). This method 
is quite successful at removing concerns about the correspondence 
between the definition of a construct and its measures. However, such 



13 
an approach leaves something to be desired in the construction of a 
theory. 

One use of "consumer satisfaction" that has dropped from vogue is 
as a social indicator (Hunt 1977a; Swan 1983; Withey 1977). The 
behavioral orientation of many of the researchers in CS/D has 
re-directed interest in the phenomenon from such attempts to examine 
general consumer welfare to elaborations on individual personal 
experience and behavior. Although the need to measure consumer welfare 
still exists, the satisfaction construct and the theory surrounding it 
have complicated the "Are consumers sufficiently satisfied?" question. 
Thus, the issue of how best to aggregate satisfaction across a 
population will not be dealt with here (see Pfaff 1977 for a discussion 
of some of these issues). 

Another controversy that arose from the desire to optimize 
consumer welfare concerned whether to define the goal as "maximizing 
satisfaction" or "minimizing dissatisfaction" (Leavitt 1976, 1977). 
Although these policy questions are not discussed in the dissertation, 
the issue of whether or not to separate satisfaction from 
dissatisfaction def initionally and theoretically is a valid concern. 

The so-called two-factor theory (attributed to Herzberg, Mausner 
and Snyderman 1959) is a direct outgrowth of the job satisfaction 
literature and is detailed by Leavitt (1977). In essence, it is felt 
that satisfaction is the result of some critical incident or process 
involving "intrinsic" factors whereas dissatisfaction results from 
"extrinsic" factors. In the job satisfaction literature, intrinsic 
factors included such attributes as achievement, recognition, 



advancement and responsibility. Extrinsic factors, on the other hand, 
are such things as salary, interpersonal relations with superiors, 
subordinates and peers, status and job security. A particularly 
salient experience involving one of these factors evokes the 
satisfaction or dissatisfaction response. In investigations of 
satisfaction, descriptions of such "critical incidents" are used to 
assess the individual's overall response. 

The difficulty with applying this rationale to consumer 
satisfaction lies in defining the types of factors a priori. That is, 
the discriminations made in the job satisfaction literature do not map 
well onto consumer situations. Leavitt attempted to classify factors 
affecting CS/D as intrinsic or extrinsic based on surface similarity to 
the classification of factors in the job satisfaction literature. The 
research failed to generate the unique pattern of influences that were 
postulated. He would have been better served by relying on or 
extending the more theoretically driven distinctions used by Swan and 
Combs (1976). 

In their paper, Swan and Combs propose that satisfaction results 
from evaluations of expressive (nonmaterial, psychological) dimensions 
and dissatisfaction results from instrumental (physical) dimensions of 
a product. They theorized that satisfaction can result only when both 
instrumental and expressive performance expectations are fulfilled. 
They admitted to difficulty in categorizing some dimensions (e.g., 
comfort) in their preliminary study involving clothing. Although they 
found some support for their operationalization of two-factor theory, 
it can be argued that the findings were bound to the critical incident 



15 



methodology. They extended the use of this technique in which people 
discuss specific occurrences that they recall to the consumer arena, 
but without arguing why these most salient events are necessarily 
representative of the majority of consumer satisfaction or 
dissatisfaction responses. 

In order to support clearly a two-factor theory of consumer 
satisfaction, better conceptual definitions from which unambiguous 
operational definitions of the two types of factors can be derived must 
be provided. Further, factors defined in this manner must be shown to 
be uniquely linked to the feeling (satisfaction or dissatisfaction) 
which is said to result. In the final analysis, Swan and Combs did not 
go beyond the disconf irmation of expectations paradigm except to order 
individual levels of expectations from physical to psychological needs 
(similar to the hierarchy attributed to Maslow 1943). This 
interpretation mirrors that by Scherf (1974) in which he applied 
Alderfer's hierarchy to consumer satisfaction. 

In an attempt to replicate the Swan and Combs findings, Maddox 
(1981) reported a similar discomfort with the discrimination between 
types of factors and the difficulties that arose in the attempt to 
operationalize the distinction. His research found different results 
from intrinsic and extrinsic factors across products. He concluded 
that the interaction with product class is probably more damaging to 
two-factor theory than supportive. 

The greatest difficulty that this perspective encounters is with 
justification as a theory versus simply being a measurement device. 
The conf irmation/disconf irmation evaluative process appears to remain 



16 



intact with both satisfiers and dissatisf iers ; no new process is 
proposed. The most likely result of this particular stream of research 
will be to identify what it is that makes the critical discrepancy 
(i.e., between expectations and perceptions) salient. In some product 
classes, it is conceivable that expressive dimensions will be more 
noticeable than instrumental dimensions when each exceeds expectations. 
For example, in the Swan and Combs case, compliments from friends would 
be more noticeable and lead to more positive responses than 
"additional" durability in clothing of college students. Similarly, 
lack of durability — fading, tears — would be more noticeable than lack 
of negative comments (assuming that students are not given to brutal 
frankness, whereas the reverse may be true if mothers were asked about 
critical incidents regarding the clothing of young children) . This 
notion of salience of the expectations-perception discrepancy will be 
discussed in Chapter 3. 

Westbrook, using similar reasoning, reports different good and bad 
emotional dimensions arising within the same process (1987). His 
measures, however, do not reduce necessarily to definitions of separate 
satisfaction and dissatisfaction constructs. Even having accepted 
dissatisfaction/satisfaction as a single continuum (that will, 
henceforth, usually be referred to simply as satisfaction), we must 
still distinguish among the cognitive, attitudinal and emotional 
definitions of the construct. Again, these must be evaluated in light 
of the underlying theory of satisfaction that each definition 
represents . 



17 



The disconf irmation of expectations paradigm (hereafter, the 
disconf irmation paradigm), which will be discussed in more detail 
presently, posits a relationship between (1) the discrepancy between 
expectations and perception and (2) certain resultant attitudes and 
behaviors. 

Some authors have defined satisfaction to be this discrepancy — or 
the judgment which identifies the discrepancy — rather than a separate 
construct (Aiello, Czepiel and Rosenberg 1977; Day 1980a). Note that 
the concern here is with the equivalence by definition of satisfaction 
and the discrepancy judgment, not with the use of one to predict (or 
measure) the other. This definition is rejected in the belief that 
satisfaction and the discrepancy judgment are conceptually distinct, 
the former carrying more motivational implications, and should remain 
separate in the lexicon derived in the study of postpurchase processes 
(Oliver and Westbrook 1982; Woodruff, Cadotte and Jenkins 1983b). 

The studies that initiated interest in postpurchase consumer 
behavior were concerned with perceptions of product performance 
(Anderson 1973; Cardozo 1965). Many marketing practitioners appear to 
equate performance with satisfaction (e.g., Ford Motor Company's 
"Number One in Customer Satisfaction" advertising campaign based on 
reports of the service requirements of new cars.) But this point of 
view ignores the potential for divergence under situations posited by 
the disconf irmation paradigm. For example, a consumer who experiences 
product performance which would be evaluated as merely fair on some 
absolute scale (indicating a low positive or even negative attitude) 
may in fact be satisfied because the performance exceeded expectations. 



(A realistic example might be a low- income consumer who has purchased 
an inexpensive used automobile. Although the car does need occasional 
repair and is not given to quick starts and high speeds, it performs 
reliably and lasts beyond its expected life.) Thus, it is probably 
best to maintain the conceptual distinction between "performance 
perception" and "satisfaction" and not attempt to implement a 
satisfaction strategy via performance engineering. (In the Ford case, 
improved performance may have been matched or even exceeded by 
expectations heightened by the promotional campaign.) A similar 
discrimination is attempted by Bahr (1982) although he makes the 
distinction using the terms "dissatisfaction" and "disappointment" 
— adding yet another concept to the literature. Another concept 
distinct from but related to satisfaction is attitude. Again there is 
a tendency to confound, by measurement devices and by definitions, 
constructs that are separate in theory (see Hartley 1967 for an 
interesting discussion of such definitional fallacies). Some authors 
have recognized this inconsistency and characterize satisfaction as "a 
special kind of attitude" that is limited to a post-purchase, 
post-consumption existence (Czepiel and Rosenberg 1977; LaTour and Peat 
1979). 

Without complicating matters any further, satisfaction can be 
maintained as a separate concept from attitude by acknowledging the 
differences in motivation level (Weiner 1985) that each implies and the 
likelihood of changes over time. LaBarbera and Mazursky (1983) cite 
Oliver's (1981) contention that the surprise or excitement elements of 
satisfaction are of finite duration; as these decay, satisfaction 



19 



decays into attitude. In view of this argument, however, it was not 
clear whether they measured attitude or satisfaction in their study. 

The relationship between satisfaction and attitude that was 
hypothesized by Oliver is compelling and is examined further in Chapter 
3. This conceptualization, however, requires definitions that allow 
one to discriminate between satisfaction and attitude and order them 
within a theoretical network. Moreover, Oliver appears to require 
satisfaction to be a construct which intervenes between emotion and 
attitude, "it is the summary psychological state resulting when the 
emotion surrounding disconfirmed expectations coupled with the 
consumer's prior feelings about the consumption experience . . . 
satisfaction soon decays into . . . one's overall attitude toward 
purchasing products" (1981, p. 27). If we accept the definition of 
attitude as "a learned implicit response that varies in intensity and 
tends to guide an individual's overt responses" (Shaw and Costanzo 
1982, p. 285), then we are left to identify the distinguishing 
characteristics of the satisfaction construct. Swan (1983) makes some 
advances by tying cognitive and affective elements of satisfaction 
together and arguing that there are potential differences in the type 
of information used to derive satisfaction than that used in deriving 
(or learning) attitude. 

Taken as a whole, the foregoing arguments sum to a 
conceptualization of satisfaction as something that mediates between a 
disconf irmation judgment and some post-experience attitude (Oliver 
1977; Oliver and Westbrook 1982). Cardozo's (1965) original concept 
also proposed that satisfaction led to or at least had distinct 



20 



influence upon particular behaviors. What is required then, is a more 

dynamic and more motivational definition of the construct than that 

provided for attitude. 

Such a definition has been proposed by Day: 

an emotional response manifested in feelings and is 
conceptually distinct from cognitive responses, brand affect 
and behavioral responses (1983, p. 113). 

Thus, Day explicitly defines satisfaction as emotion rather than a 
precursor (as with perceptual or attitudinal definitions) or a product 
(as Oliver 1981 describes). Chapter 3 introduces and discusses 
theories of emotion and their relationship to more general motivation. 

The emotional character of this definition is amplified by 
developmental work conducted by Westbrook (1983, 1987). He used 
measures derived from Izard's Differential Emotions Scale (1972) to 
describe emotional profiles of post-purchase consumers. (This theory 
of emotions is explicated further in Chapter 3.) By defining 
satisfaction as an emotional phenomenon and using measures of emotion 
in its assessment, Westbrook was better able to describe the construct 
directly, albeit as a result of emotion (1987). Moreover, as a concept 
founded on emotions, satisfaction is unique in the marketing 
literature. Thus, the overlap between related constructs, such as 
disconf irmation and attitude, is eliminated and the potential for 
measuring each individually is enhanced. 

The adopted definition should also be held to be distinct from one 
final class of definitions: the (cognitive) evaluation of emotions (Day 
1980a; Oliver 1977; Westbrook 1981). This has been characterized as a 
"stepping away from an experience and evaluating it" (Hunt 1977a, 



21 



p. 459). This formulation seems to require that emotion precede 
cognition (and, presumably, the behaviors that have been posited). 
Whether or not other cognitions precede these emotions is not clear. 
These definitions have not been pursued vigorously in the CS/D 
literature (their proponents having moved to more emotional and less 
cognitive versions) and are therefore also rejected in this 
dissertation. 

In summary, it is perhaps best to recall the argument raised by 
Swan (1983) that the question of what satisfaction is "in essence" is 
not answerable. Rather, we wish to achieve a consensus accepting a 
rigorous definition that is useful in a larger theoretical framework/ 
an approach that is more consistent with a functionalist approach 
(Alderson 1965). These requirements are met by Day's definition 
presented above and the framework and measurement techniques presented 
below. 

The Disconf irmation Paradigm 

As was stated previously, the one major agreement within CS/D has 
been the acceptance of the disconf irmation paradigm. Essentially all 
of the theory and research in the area derives in part from an assumed 
interaction between expectations and perceived outcomes, and a strong 
natural preference for the expected over the unexpected (Silverman 
1968). 

This recognition of the crucial role of expectations traces to the 
work by Cardozo (1965). He introduced to marketing the notion that 
prior expectations may influence one's perception of reality (Aronson 
and Carlsmith 1962; Watts 1968). Prior to this, physical performance 



22 



of a product was thought to drive directly consumer response. Cardozo 
suggested that expectations moderate the perception of this physical 
performance. It is this modified perception, which he called 
satisfaction, that drives the post-purchase responses (Figure 1 ) . For 
a time, interest in the post-purchase consumer continued to focus on 
this interplay between expectations and perception (generally, 
perceived product performance). The next major study in the area 
attributes the conf irmation/disconf irmation episode to the decision 
made by the purchaser rather than to a simple confirmation of 
particular expectations (Cohen and Goldberg 1970). This formulation is 
particularly useful because it avoids the complications that arise from 
attempts to specify whether expectations and performance react at an 
attribute specific or summary level. This issue and others raised by 
Cohen and Goldberg which have not as yet been pursued in CS/D are 
discussed in more detail in Chapter 3. 

Anderson and Hair (1972) provided an interesting and useful 
perspective on the role of expectations. They argued that consumer 
satisfaction decreases as the consumer's experience with products of 
improving quality increases because improvements in quality raise 
expectations. They also point out the potentially self-defeating 
nature of promotion--i . e . , the possibility of raising expectations too 
high — which was alluded to in Chapter 1 . In their paper, consumer 
satisfaction and product perception were treated as conceptually 
distinct (Figure 2). Expectations were still thought to affect product 
perception (through the same sort of mechanism that Cardozo had 
proposed), and this relationship became the focus of the empirical 



23 



Expectations 



Physical 
Performance 



Perceived 
-> Performance — 
"Satisfaction' 



Repeat Purchase 
•> Product Line Acceptance 
Word-of-Mouth 



Figure 1 
Cardozo's Model of Satisfaction 



based on Cardozo (1965) 



24 



Expectations 



Product -- 
(Objective) 



Product 
Perception 



> Satisfaction 

Dissatisfaction 



Figure 2 
Anderson and Hair Treatment of Satisfaction 



Based on Anderson and Hair 1972 



25 
study. The distinction between perception and satisfaction was lost in 
the study design. 

Anderson restates the conceptualization in his later, more 
frequently cited paper (1973), in which he proposes that although 
satisfaction has no literal definition it "might be measured by the 
degree of disparity between expectations and perceived product 
performance" (p. 38). Operationally, however, he still treats perceived 
performance as the variable of interest and does not further elaborate 
the relationship with satisfaction. Other investigations of 
post-purchase evaluations have not invoked the satisfaction construct 
and thus do not perpetuate the ambiguity in terms (Oliver 1977; 
Olshavsky and Miller 1972; Winter 1974), although each is occasionally 
cited as having been a satisfaction study. 

LaTour and Peat (1979) reviewed this "product performance" 
literature and were careful to point out that "perceived product 
performance cr quality is not necessarily the same as satisfaction" 
(p. 432). They detailed three psychological explanations of how 
expectations and reality may interact to produce perceived performance: 
1) contrast; 2) assimilation-contrast; and 3) cognitive consistency 
theories. They conclude that expectations are really only likely to 
affect perception when objective performance is ambiguous or otherwise 
difficult to assess (for example, when dealing with providers of 
services; Zeithaml 1981). Such modifications, when they occur, would 
resemble an assimilation effect (i.e., perception would be closer to 
expectations than it would in objective reality) and would presumably 
be driven by the assumed value of the information provided by prior 



26 



expectations versus that provided by the physical product. This is 
explicated more fully in their presentation of a comparison level 
theory of satisfaction discussed below (see also LaTour and Peat 1978; 
1980). 

LaTour and Peat conceived of satisfaction as equivalent to 
post-consumption attitude. This is incompatible both with the 
orientation taken in the dissertation and with more recent 
comprehensive satisfaction models that include attitude as a separate 
(but related) construct. Again, the most compelling of these, and the 
one on which the dissertation builds, was proposed by Day (1983) 
(Figure 3) . 

In this model, perception, disconf irmation, attitude (affect) and 
the satisfaction feeling are separate constructs. The comparison of 
perception to expectation may result in positive or negative 
disconf irmation or in confirmation of the expectation. 1 The 
disconf irmation judgment results in the feeling that has been defined 
as "satisfaction." This satisfaction feeling, in turn, drives 
specific responses, including word-of-mouth or complaining, changes in 
repurchase intention for the brand and/or the product class. 
Satisfaction feelings also influence brand and/or product class 
attitude (which Day called affect) after the evaluation has occurred. 
As presented, this is not intended to be a causal model because some 
relationships may be more important for certain product-usage 
situations than others (e.g., a link between expectations and 



1 For simplicity, the comparison and its result are hereafter 
referred to as disconf irmation. 



m 






c 













•H 




CO 


4J 




e 


(0 




m 


-U 




o 


u 




Z 


QJ 






a 






X 






w 







CO 

0) 
CO 

c 
o 
a 

CO 
0) 

<X 



27 



O CO u 

a co id 

>tj (0 4-1 

u u < 



ZET 



"v. c 

c o 

O -H 

•H 4J 

4J <0 

«3 E 

e u 



4-1 C 

c o 

O u 

U CO 



c 
o 

■H 

0) 

u 



T3 

C <D 

(0 4-1 

M 4-1 

CQ < 



3T 



u co u 

3 CO d) 
O "-I 4-1 



T3 CJ 

C 0) 

rc 4-1 

M 4-1 

CQ < 






ai 




en 


U T3 


O 


<D 


-H 


<-\ 


M 


3 


Cm 


O 




c 




* 



u 

(0 

4-4 

CD CO 
U -H 
3 -P 
CD (0 

•h cn 

CD 

X! 



n 



a 

e 
o 

P. 

4-1 

T3 
Q) 



•a 

o 



28 



perception as just discussed). What the figure does provide, however, 
is a diagram of how the flow would proceed in the prototypical 
situation. 

Day has taken a crucial step in developing the CS/D literature. 
In attempting to derive a consensus, he has proposed a framework for 
all consumer satisfaction which could be restated as a set of lawlike 
generalizations, each of which can form the basis of sub-theories. 
Relationships posited here that have been examined (e.g., 
disconf irmation causes satisfaction versus perception causes 
satisfaction) have not been presented as part of a grander theory or 
nomological network (S. Hunt 1976). The increased level of generality 
that is represented by Day's model is carried further by the 
conceptualization of a general satisfaction construct that is 
explicated in Chapter 3 (note that, even for Day's model, no clear 
discrimination is made between product class level and brand level 
constructs) . 

Other comprehensive satisfaction theories have recently been 
proposed (Table 1). Woodruff, Cadotte and Jenkins (1983a) describe a 
model very similar to Day's. They, like Day, separate norms from 
expectations and define satisfaction as an emotion. A key element in 
this model is the attempt to specify the relationship between the 
cognitive disconf irmation process and the emotional result 
(satisfaction). Specifically, they predict that negative 
disconf irmation (performance falls below norms) will lead to negative 
feelings (dissatisfaction); positive disconf irmation (performance 
exceeds norms) will lead to positive feelings (satisfaction); and 



29 



Table 1 
Perspectives on Consumer Satisfaction 



Paradigm 
Disconf irmation 



Variants 

two factor 



mood 



experience 
(also 

comparison 
level and 
value percept 
disparity) 



Dependent variables 

immediate postpurchase 
satisfaction, intentions, 
repurchase 

postpurchase satisfaction 



postpurchase satisfaction 



postpurchase satisfaction 



indifference postpurchase satisfaction 



attribution 



Equity 



Normative 
deficit 



Composite 



postpurchase satisfaction, 
complaining 



distress or postpurchase 
satisfaction 



satisfaction with goods 
and services as comprising 
a standard of living 

overall (global) 
satisfaction 



Causal mechanism 

Satisfaction covaries with 
difference between 
performance and expectations 

satisfaction results from 
performance on instrumental 
dimensions, dissatisfaction 
results form performance on 
expressive dimensions 

satisfaction results in part 
from mood state (or business 
attitude) at purchase 

expectations are discrete 
from post product experience, 
both affect satisfaction, 
expectations derive from 
many sources 



satisfaction occurs under 
usual disconf irmation 
conditions but no such 
feeling results when 
confirmation occurs 

problems attributed to 
external (i.e., marketer- 
controlled) sources are more 
likely to arouse 
dissatisfaction and 
complaints 

distress increases as 
consumer's outcome becomes 
less favorable than (1) the 
marketer's or (2) another 
consumer' s 

satisfaction occurs when the 
quality of goods and services 
equals social norms 

overall satisfaction = 
summation of satisfaction 
with attributes 



Note: Based on Swan (1983) 



30 
simple confirmation (performance functionally equivalent to norms) will 
lead to a neutral result. 

The perspective offered by Woodruff et al. is rejected on a number 
of grounds. The limiting of expectations to brand specific predictions 
of performance is unsupported in the CS/D literature and an unnecessary 
complication. Rather than separate brand and product prior experience, 
it is more parsimonious to consider these as separate influences on a 
single expectation construct. Similarly, although the suggestion that 
imperfect perceptions and comparison processes lead to discrete, step 
function disconf irmation judgments and resultant satisfaction feelings 
does provide a methodological advance over the previously assumed 
continuous relationships, this is not a development specific to 
satisfaction theory. Rather, Woodruff et al. have proposed technical 
approaches to examining perceptual discrimination of performance or 
performance versus expectation discrepancies (following a kind of 
Signal Detection Theory logic) but no real modification to the 
disconf irmation paradigm. 

Other authors have sought to evaluate versions of the 
disconf irmation paradigm through causal modeling. Churchill and 
Surprenant (1982) report evidence of different processes in the cases 
of satisfaction with a nondurable (chrysanthemum) and a durable product 
(video disc player), with the former depending on both performance and 
disconf irmation, and the latter dependent only on product performance. 
In a similar vein but with contradictory results, Wilton and Tse (1983) 
report that satisfaction with a compact disk 



31 



player (a durable product) was dependent on both performance and 
dis confirmation. 

While the intent of each of these studies (i.e., testing the 
complete process model) is laudable, it can be argued that both suffer 
from conceptual and methodological defects that render the findings 
equivocal. At this point it should suffice to note that neither 
experiment operationalized a purchase situation. Because both are 
interested in consumer satisfaction and both define this construct as 
post-purchase , they possess a fatal flaw. This issue and others are 
discussed in more detail in Chapter 3 which presents a more 
comprehensive description of the prototypical satisfaction evoking 
situation and discusses the conditions necessary to evoke the feeling. 

Finally, some modifications to the disconf irmation paradigm have 
been proposed in the CS/D literature that are beyond the scope of 
Day's conceptualization of satisfaction and are not central to the 
dissertation. These embellishments do however reflect how the general 
paradigm can be elaborated by detailed study within certain portions of 
the model. 

The first of these has already been alluded to; that is, the 
possibility that a "zone of indifference" exists which may result in no 
particular satisfaction feeling resulting from a given consumer 
situation. Woodruff, Cadotte and Jenkins (1983a) and Swan and Trawick 
(1980a) assert that satisfaction feelings do not result when 
performance simply meets expectations (or the discrepancy between them 
is not detectable) and include this confirmation effect as part of the 
theory. Both Dav (1977) and the current conceptualization include a 



32 



neutral state or lack of satisfaction feeling as a potential outcome; 
however, it is not felt to be specifically linked to a confirmatory 
judgment (i.e., a lack of a discrepancy). 

Other authors have investigated the influence of general attitudes 
toward business (Anderson, Engledow and Becker 1979) or the consumer's 
overall emotional state or mood (Westbrook and Cote 1980) on the 
satisfaction process. Although each reports support for relationships 
in the directions hypothesized, it is conceivable that the dependent 
variable in each study was attitude rather than satisfaction. While it 
is not argued that attitudes toward business and/or mood will not 
influence satisfaction, neither construct is central to the process. 

As mentioned above in the section concerning definitions of 
satisfaction, the Two-factor Theory (Swan and Combs 1976; Swan and 
Trawick 1980b) can be considered an extension of the basic 
disconf irmation paradigm. Essentially, Swan and his colleagues (and 
others--see Day 1983) have chosen to focus attention on the 
expectations portion of the model. Although presented as separate 
constructs affecting satisfaction, each of the variables proposed can 
be considered to be an influence on the satisfaction construct. That 
is, the expectations that become important in the disconf irmation 
experience may reflect attitude toward business, mood, or only certain 
physical product dimensions (categorized as instrumental or 
expressive), depending on the situation. 

Finally, the CS/D literature is so young that the investigation of 
a single end-user's satisfaction with consumer products is sufficiently 
complex to challenge the field. Lambert, Dornoff and Kernan (1977), 



33 



however, open the way to more expanded horizons by looking at the 
post-purchase industrial buyer. Although only an exploratory study 
involving post-choice attitudes, the paper does raise the question of 
whether or not the disconf irmation paradigm would be amenable to 
applications in which buying centers or other group decision-makers are 
involved as well as situations in which the buyer is not the user. 
Given the perspective involving emotions that this dissertation has 
adopted, it is difficult to address these situations within the 
classical satisfaction framework. It is argued in Chapter 3 that, 
among other causes, satisfaction emotions result from the decision to 
commit to and personal experience with the product. Thus, the 
situations suggested by Lambert et al. do not translate directly into 
the satisfaction paradigm (note that they do not invoke this paradigm — 
they are interested in attitudes after a choice is made) . 
Other Perspectives on Satisfaction 

Although the disconf irmation paradigm dominates the CS/D 
literature, some theorists have proposed other views purportedly 
distinct from this perspective. Swan (1983) briefly summarizes three 
paradigms in addition to disconf irmation. Of the three, only the 
equity approach considers specific satisfaction with a good or service 
at the same, summary, level as the current form of the disconf irmation 
paradigm. 

Equity theory, borrowed from social psychology, basically posits 
that a person involved in an exchange relationship compares the ratio 
of his outcomes from : inputs to the exchange with a similar ratio for 
another person involved in the relationship. When the ratios are 



34 



equal, the exchange is said to be equitable; when unequal, the exchange 
is inequitable. Inequitable relationships create distress which 
results in attempts by the individuals to restore equity (Shaw and 
Costanzo 1982). 

Huppertz (1979) introduced equity theory into the CS/D literature, 
followed by Mowen and Grove (1983) and Evans (1983) (see also Huppertz, 
Arenson and Evans 1978). None of the authors has dealt with the 
critical conceptual issue of specifying how the comparison other will 
be identified. The theory, as presented in social psychology, 
generally assumes that the comparison other is the one with whom the 
exchange relationship is forged (Shaw and Costanzo 1982--e.g., a retail 
merchant). Mowen and Grove used another consumer who got a better or 
worse deal in a similar exchange as the comparison in order to generate 
distress; Huppertz et al. studied memorable exchanges subjects had 
actually experienced with merchants. 

Another problem faced by equity theory analyses, even in social 
psychology, is the quantification of the inputs and outcomes of each 
party. Even if this can be handled in devising an experimental 
protocol, the issue of how the evaluation process operates in the 
typical consumer situation remains unexplicated. In summary, 
application of equity theory to satisfaction is still in its infancy, 
as suggested by Swan (1983). Before this approach will gain 
acceptance, its particular conceptual questions must be addressed along 
with the issues of definition and measurement with which the 
disconf irmation paradigm is still wrestling. Evans, for example, from 



35 



the perspective taken here conducted an attitude study, rather than a 
true satisfaction study. 

Swan discussed a second paradigm that he labelled the Composite 
approach. This perspective does not posit a process different from 
disconf irmation, but suggests that the judgment occurs at attribute 
specific levels and satisfaction with the product is a composite of 
these separate satisfactions. As an example, Swan cites Westbrook's 
(1981) investigation which explained department store satisfaction "in 
terms of satisfaction with sales personnel, store appearance, 
merchandise selection, etc." (1983, p. 124). 

Further development of Composite approaches is awaited. For 
example, it would be useful to understand how one might discriminate 
between a "summing" of discrepancy cognitions and a summing of 
satisfactions, each resulting in an overall satisfaction. In 
particular, the summing of emotions (as satisfaction is defined in 
Chapters 3 and 5) is a difficult concept to reconcile. The next 
section of the present chapter builds its discussion of Types of 
Satisfaction on a similar hierarchical logic. However, discriminations 
are at the expectation and perception levels and the conceptualization 
remains within the d iscc-nf irmation paradigm. 

The last paradigm, the Normative Deficit approach, is more 
concerned with the qualitv of life definition of satisfaction that was 
dismissed earlier in this chapter (Morris, Winter and Beutler 1976). 
This perspective assumes that societies possess norms regarding what an 
individual or household should expect from the marketplace. To the 
extent that one's current state does not reach this level, 



56 



dissatisfaction occurs. Swan uses the example of housing status that 
requires older children of opposite sexes to share a bedroom. As 
outlined, Normative Deficit shares some concepts with the Mowen and 
Grove operationalization of equity theory, i.e., "I should get what the 
other guy is getting." This view, however is less concerned with 
relative comparisons (inputs versus outcomes) than with absolute 
comparisons (specific ownership per individual or household) (Morris 
1977; Withey 1977). This version of consumer satisfaction may be 
better captured in other terms such as "anomie" (Stein 1980) or 
"consumer discontent" (Lundstrom and Lamont 1976), that attempt to 
describe the totality of the consumer X marketplace interaction. It is 
not clear, for example, from Swan's discussion whether a Normative 
Surplus would create satisfaction beyond that achieved by Normative 
Parity. 

Approaches to satisfaction other than those discussed by * Swan may 
be considered extra-paradigmmatic by some in the field. Two-factor 
theory (Maddox 1981; Swan and Combs 1976) has already been determined 
to be more a theory of expectations than a theory of satisfaction. 
Similarly, Comparison level theories (Barbeau and Quails 1984; LaTour 
and Peat 1978, 1979, 1980) are also concerned with specifying what it 
is that is compared against absolute performance. These too, can be 
considered expectations under the disconf irmation paradigm (see the 
section in Chapter 3 on Expectations). Other models have been proposed 
and discussed as "alternatives" to the disconf irmation paradigm. These 
generally have been theories borrowed from Psychology and applied to 
expectation formation. Principal among these are applications of the 



37 



Brunswik Lens model (Tapp 1984) and Value percept disparity; i.e., 
conformity with a person's value also influences satisfaction 
(Westbrook and Reilly 1983). As expectation theories, these can be 
subsumed within the disconf irmation framework. 2 

Attribution theory has been introduced into the CS/D literature as 
an explanation for consumers' choice of alternative responses to a 
dissatisfying experience (Folkes 1984a, 1984b; Valle and Wallendorf 
1977). In this context, it is really a part of the consumer 
complaining literature which is discussed briefly below in the section 
concerning Results of Satisfaction. 

More recent theory and research by Folkes and her colleagues 
(Folkes, Koletsky and Graham 1987; Folkes and Kotsos 1986) have 
addressed the influence of attribution on satisfaction itself. They, 
and others (e.g., Westbrook 1987), argue that the source of 
disconf irmation influences the arousal of emotions. This formulation 
reflects the concept of locus of control or responsibility (Carlsmith 
and Freedman 1968; Weiner, Russell and Lerman 1978) that posits 
different responses to disconf irmation attributable to one's self, 
another party or an uncontrolled situation. Westbrook describes the 
likely emotional profile that would result in each instance. 

Oliva and Burns (1978) present an innovative mathematical 
approach, Catastrophe Theory, to discriminate between those who will 
and those who will not engage in complaining behavior. Although 



2 Another theory, constraint-reinforcement (Shelly 1972) reflects 
more of an operant conditioning approach to the concept. This 
behaviorist tradition is sufficiently far afield of the other theories 
discussed that it is not investigated further. 



38 



neither a psychological approach nor even a satisfaction paper per se, 
this work is mentioned here because the mathematics proposed could 
present an interesting way of discriminating satisfied from 
dissatisfied consumers. This theory has some promise in integrating 
attempts to sum attributes (Composite approach) with those separating 
attributes into categories (Two-factor theory) and with the more 
dominant, gestalt approaches. In essence, the theory offers a 
mathematical interpretation of discontinuous phenomena that are 
dependent on direction (e.g., it may be a greater distance from left to 
right than vice versa). The mathematics may provide solutions to such 
problems as the apparently uncertain levels of disconf irmation that are 
necessary to reach "satisfaction" or to prompt complaining. The 
further application of such mathematics, by suitably sophisticated 
users, could provide general models to define such states. As 
indicated before, the CS/D literature is sorely in need of basic 
generalizations . 

Types of Consumer Satis faction 

CS/D theories or paradigms have developed predicated on the 
assumption that "consumer satisfaction" was the construct under 
scrutiny. In this section, the various levels of abstraction at which 
the concept has been operationalized are discussed and the necessity of 
specifying the relationships among these is demonstrated through a 
rationale derived from the literature that examines the consistency 
between attitude and behavior. 

Consumer satisfaction is a term that has been applied to 
consumers' responses to target stimuli ranging from a particular 



39 



product usage occasion to the totality of life (Table 2). Of these, 
the most often discussed and operationalized in the literature has been 
satisfaction with a particular item, generally a product (Westbrook 
1980c). This is perhaps a vestige of surveys that had focused on 
individual items that prompted a satisfaction (usually negative) 
response that was of interest to marketers (e.g. Day and Landon 1976; 
Summers and Granbois 1977). An example of Individual Item 
satisfaction can be drawn from one condition of a study reported by 
Churchill and Surprenant (1982; Surprenant and Churchill 1984). The 
study concerned modelling the satisfaction process for two types of 
goods: durables (a video disc player) and nondurabies (a chrysanthemum 
plant). Subjects in one condition of the design were given a 
chrysanthemum and interviewed as to their satisfaction with the plant 
after four weeks. Thus, they were exposed to this one example of the 
new hybrid and the questioning focused on the plant they had been 
given, and on the experience they had with it over the four week 
"trial" period. 

In contrast, a corresponding group of subjects was merely shown a 
four-week old plant and asked to respond to a similar questionnaire. 3 
This latter condition mirrors the designs of the early product 
performance studies (Anderson 1973; Cardozo 1965) in which only a 
single experience with a single product was allowed. This roughly 
corresponds to satisfaction with the Purchase/Usage Occasion, the 



3 Problems with the design of this study and implications for its 
usefulness as an examination of satisfaction as defined here are 
discussed in Chapter 3. At issue presently is simply the level of 
abstraction of the target of the satisfaction feeling. 



40 



Table 2 
Types of Consumer Satisfaction 



Type 



Cited in 



Purchase/Usage Occasion 
Individual Item 

Brand/Manufacturer 

Product Class 

Store/Retailer 

Marketplace (Consumer 
Discontent) 



(Surprenant and Churchill 1984) 

(Churchill and Surprenant 1982; Shuptrine 
& Wenglorz 1980) 

(Diener 1977) 

(Gibson 1981; Hunt 1976; Ortinau 1982) 

(Westbrook 1981 ) 



(Aiello and Czepiel 1979; Lundstrom and 
Lamont 1976; Hunt 1976) 



Life 



(Sherman and Schiffman 1984; Withey 1977) 



41 



classification which heads Table 2. In many contexts, satisfaction 
with the Item is resultant from satisfaction with the series of Usage 
Occasions involving it. 

To a degree, the satisfaction types in Table 2 are arrayed along 
this dimension of cumulativeness or generality (although this is not 
strictly true). Specifying the exact relationships among these is well 
beyond the scope of the dissertation but two varieties of relationship 
are suggested. As just discussed, a number of more specific responses 
may accumulate in some way (additive, Bayesian, random walk?) and 
generate a more general response (Andreasen 1977a, b). On the other 
hand, a specific stimulus may be interpreted as representative of the 
more general stimuli (e.g., the use obtained on this sample Occasion is 
judged to be typical for this particular Item) and the response will be 
attributed to the broader classification (using this Item will always 
make me feel this good/bad). This latter prediction is based on the 
same rationale that predicts that more general norms (e.g. how this 
Item performed in the past) influence a more specific expectation (how 
this Item will perform on this Usage Occasion). Similarly, users may 
choose to use the Occasion as case information in order to make 
judgments about the Item, as discussed in the decision literature (Ofir 
1982). 

The same logic carries through to generalizing beyond satisfaction 
with the Individual Item. Whereas Westbrook (1987; 1933; 1980a) 
examines consumers' responses to their individual automobile, Ortinau 
(1982) asks consumers to use personal experience with their recently- 
purchased automobile in responding to questions about the model 



42 



(Pontiac Grand Prix) in general. By similar extrapolation, one might 
wish to consider satisfaction with the entire product category (Hunt 
1976; Shuptrine and Wenglorz 1980), which may later result in more 
generalized behaviors (e.g. commenting to a friend about American cars 
or automobiles as a whole). 

Similarly, a satisfaction response to the Item can result in 
generalization to the Manufacturer or Brand even though they are no 
longer responsible for the product. Diener (1977) reports negative 
attributions to manufacturers of cosmetic products that were 
re-packaged and re-sold as samples by an unscrupulous and unauthorized 
vendor. In this particular case, the actual retailer was virtually 
unidentified and did not become the target of dissatisfaction feelings 
or responses. 

Satisfaction with the Retailer has, however, been the subject of 
some CS/D research. Westbrook (1981) discusses Retailer satisfaction 
as resulting from a set of component satisfactions (e.g., with 
salespersons, store environment, etc.) that distinguish it from Product 
satisfaction (see also Rodgers and Sweeney 1980). Other authors have 
discussed Retailer satisfaction implicitly as resulting from 
satisfaction with products carried (e.g., Bearden and Teel 1983) or 
follow-up service (Vredenburg and Wee 1986). This conceptualization 
has also been extended by authors interested in complaints about 
products and how consumers' satisfaction with complaint handling 
affects satisfaction with the Retailer (Diener 1980; Gilly and Gelb 
1982; Lippert and MacDonald 1981; Resnik, Gnauck and Aldrich 1977; 
Resnik and Harmon 1983; Summers and Granbois 1977). 



43 



Aiello and Czepiel (1979) posit a similar hierarchy in which 
Product/Service satisfaction generalizes to Enterprise Satisfaction 
which similarly generalizes to System satisfaction (see also Aiello, 
Czepiel and Rosenberg 1977). This last concept, referring to 
"consumers' subjective evaluation of the total benefits received from 
the operations of the institutional marketing system" (Aiello and 
Czepiel 1979, p. 129), closely resembles the Consumer Discontent 
definition of consumer satisfaction discussed previously. This level 
of analysis has quite a history in the early CS/D literature (Hunt 
1976; Lundstrom and Lamont 1976; Scherf 1974; Strumpel 1974) but is of 
less concern in present theorizing (but see Liefeld 1980; Olander 
1980). 

This very generalized satisfaction with the Marketplace has also 
been implicated as one determinant of satisfaction with Life (Leavitt 
1976; Sherman and Schiffman 1984; Witney 1977). As an ultimate 
generalization of the concept, Life satisfaction also includes Job 
satisfaction and other non-consumer behavior types that are discussed 
briefly in Chapter 3. Presumably, Life satisfaction represents the 
result of a cumulative rather than an attr ibutional process as had been 
described previousiv. 

Such processes are also posited in relationships among Individual 
Item satisfaction and satisfaction with (a) the purchase process (Day 
1977; Wanous and Lawler 1972), (b) warranty service (Bernacchi, Kono 
and Willette 1930), and (c) other indirect aspects of the product such 
as a restaurant's use of consumer surveys (Swan, Trawick and Carroll 



44 



1981) or (d) the systems used to deliver health care services 
(Langmeyer and Miaoulis 1981; Locker and Dunt 1978). 
At titudinal -Behavioral Consistency 

As the discussion of types of satisfaction becomes more 
fine-grained, the necessity of specifying all constructs in a given 
model or operationalization at the same level of generality becomes 
evident. This issue has been discussed at length in the attitude 
literature. A brief review of the conclusions provides a basis for 
evaluating the usefulness of the concept of levels of specificity for 
CS/D. 

In the attitude literature, the most direct analogy to the 
preceding discussion is Fishbein's (1966) discrimination between 
attitude toward an object and attitude toward a class which includes 
that particular object. He argued that, although these may be related 
concepts, "It seems fairly obvious that our chances of predicting 
behavior from attitude are practically nil until we at least start 
measuring attitudes toward the appropriate stimulus." (p. 222). On 
this basis, he critiqued LaPiere's (1934) study that reported 
apparently contradictory attitudes and behaviors toward Orientals. 
Fishbein argued that the Oriental couple in LaPiere's study so 
mismatched the stereotypic Oriental that attitude toward Orientals 
would be essentially irrelevant (i.e. attitude toward this couple would 
dominate) in determining behavior and, therefore, the finding of no 
correlation is not counterintuitive. 4 



4 There are other grounds raised by Fishbein on which to criticize 
the LaPiere study. The argument just cited, however, remains valid 
even allowing for these other difficulties. 



45 



Fishbein extended this work in later writings. In an article 
discussing attitudinal-behavioral consistency (1972), he revised a 
prior theory (attributed to Dulanv) and introduced the notion of 
attitude toward the act as one of the determinants of action (rather 
than the more general attitude toward the object(s) affected by the 
action). Again, his argument was based upon the appropriateness of 
making predictions about specific behaviors (or behavioral intentions) 
based on specific attitudes. This is similar to the concept of 
attitude toward the situation introduced by Rokeach, "an attitude may- 
be focused on either an object or a situation . . . behavior is always 
a function of at least these two types of attitudes" (1970, p. 135). 

Finally, Fishbein and Ajzen pose a similar question about 
consistency between behavioral intention and behavior in a discussion 
of behavior toward the church, "correlations between intention and 
behavior were higher when the levels of specificity tended to 
correspond than when they did not" (1975, p. 350). It is this concept 
of levels of specificity that is most relevant for the CS/D 
literature . 

The studies discussed in the prior section each purport to 
investigate some aspect of "Consumer Satisfaction". Drawing on the 
rationale developed from the attitude literature, we can conclude that 
there is no inherent "consumer satisfaction" construct but, more 
appropriately, separate satisfactions with certain target concepts 
(e.g. particular items or services, marketing entities, etc.) that are 
part of a consumer context. Furthermore, these satisfactions are 



46 



interrelated within a level of specificity hierarchy that will enable 
predictions of effects on attitudes and behaviors. 

More concretely, consider the following example. A consumer finds 
that, on using her new calculator for the first time, the 
multiplication button fails to function. Her feeling of 
dissatisfaction on this occasion may have as its target (a) the 
particular calculator, (b) the brand of calculator, or (c) calculators 
in general (among other targets). Prediction of which type of 
satisfaction would demonstrate the greatest effects would depend on 
such factors as the consumers' specific expectations and/or attribution 
of the cause of the poor performance. Adequate measurement, then, 
requires that such specificity be built into the techniques used. 
Moreover, hypotheses concerning subsequent behaviors will also be 
required to be drawn at the same level of specificity (e.g., (a) 
exchange for a different unit of the same model, (b) change brands, or 
(c) find another alternative for doing arithmetic, respectively). 

On the basis of the foregoing discussion, the term "Consumer 
Satisfaction" is rejected as being indicative of a particular construct 
or concept. Rather, the term "satisfaction", as defined previously, 
will be used to refer to a feeling that results from or has as 
"Targets" particular stimuli. Consumer satisfaction (or CS/D), then 
refers to situations or the literature that discusses the feeling with 
respect to stimuli generally regarded as part of the consumer domain 
(e.g., products, retailers). This generalization of the satisfaction 
construct is extended beyond CS/D in the next chapter. 



47 



Measurement Issues in the CS/'D Literature 
Before introducing a new conceptualization and new measures of 
satisfaction in Chapter 3, it is useful to consider the measurement 
techniques that have been developed and discussed at some length in 
prior publications. The earliest discussions of measurement focused on 
evaluations of product performance (Anderson 1973; Miller 1977; Oliver 
1977) and on what we are now calling consumer discontent (Lundstrom and 
Lamont 1976; Pfaff 1977). Since these are now thought to comprise a 
separate construct in the satisfaction process and a more general 
attitudinal phenomenon, respectively, and are distinct from the feeling 
and theory of interest these measures are not reviewed in depth here. 

At the opposite level of specificity from attribute specific 
measures are scales intended to evaluate one's satisfaction with life 
(Andrews and Witney 1976) or the marketplace (Lundstrom and Lamont 
1976; Pfaff 1977). As constructed, these resemble the multi-attribute 
measures developed to investigate job satisfaction (e.g., Wanous and 
Lawler 1972). In the organizational behavior literature, this 
treatment of job satisfaction is intended to reflect a more enduring 
attitude than a specific, acute feeling (compared, for example, with 
Herzberg's interest in critical incidents). Similarly, Consumer 
Discontent and its measures, while useful in many contexts (e.g., 
economics, public policy questions) does not provide a viable 
alternative to the measurement techniques that are discussed 
presently. 

If we agree that perception of performance comprises a construct 
that precedes satisfaction, then it is useful to have good measures of 



48 



this. These could take the form of measures of overall value such as 
used by Anderson (1973) or measures of performance on an attribute 
specific level (Oliver 1977; Oliver and Linda 1981). The Oliver (1977) 
article does, however, suggest that caution is necessary in order to 
estimate perception as separate from disconf irmation. 
Specific Measures Use d 

Although considered to be separate from the feeling of 
satisfaction in the conceptualization adopted for the dissertation 
(Figure 3), the disconf irmation cognition does play a role and has 
historically been used as an operationalization of satisfaction. Thus 
measures of disconf irmation are included as the first entry in Figure 
4a. Figure 4 comprises a taxonomy of measures of satisfaction that 
differs somewhat from Andreasen's (1977a:b). Whereas Andreasen was 
concerned with the influence of time (which is discussed in the next 
chapter; and the relative "objectivity" of measures, the present 
taxonomy is based more on the nature of the measures themselves than on 
any implicit theory of satisfaction or theory of measurement. As such, 
two dimensions evolved: (1) a cogni tive-af f ective-conative dimension 
reflecting what is being measured, and (2) a verbal-graphic dimension 
reflecting how the measures are collected. 5 

Evaluative verbal measures of satisfaction, such as those 
presented in Figure 4a, are the most commonly used of the techniques 



5 At this point, no distinction is being made between these 
measures which treat satisfaction/dissatisfaction as a single continuum 
and those which posit separate constructs, as discussed previously. 
For the most part, the scales presented could be adapted for either 
conceptualization: they are shown in as close to original form as is 
feasible . 



49 



(a) Evaluative/Cognitive Measures in Consumer Satisfaction 

Verbal 

Disconf irmation measures 

1. My expectations were: 

Too high: Accurate: Too low: 

It was poorer It was just as It was better 

than I thought I had expected than I thought 



(Oliver 1977; 



was much better (worse)' than I expected. 

Very Strong Strong Strong Very Strong 

Yes Yes Yes ? No No No 



(Swan, Trawick and Carroll 1931) 



Much more Somewhat About Somewhat Much less 

than I more than what I less than than I 

expected I expected expected I expected expected 

1 2 3 4 5 

(Aiello, Czepiel, and Rosenberg 197 7^ 



Derived measure for attribute levels 

(Level currently provided) - (Level ideally desired) = 
Disconf irmation 

(Westbrook and Oliver 1981) 



Degree of satisfaction measures 

5. Overall, how satisfied have you been with this ? 

100% 90 80 70 60 50 40 30 20 10 0% 
Completely (Half & Not at 

Satisfied Half) all Satisfied 

(Mocre & Shuptrine 1984: Oliver & Eearden 1933; 
Oliver & Westbrook 1982; Westbrook 1980b; Westbrook 1981). 

Figure 4 
Measures Used in Consumer Satisfaction Research 



50 
was very satisfactory (unsatisfactory) 

Very Strong Strong Strong Very Strong 

Yes Yes Yes ? No No No 



(Swan, Trawick and Carroll 1981 



7. How satisfied were you with 



Very Somewhat Slightly Slightly 

Dissatisfied Dissatisfied Dissatisfied Neither Satisfied 



Somewhat Very 
Satisfied Satisfied 



(Oliver and Bearden 1983; Oliver and Linda 1981) 



8. Were you satisfied/dissatisfied? (Choose one.) 

(Gronhaug and Arndt 1980) 



I am always or I am I am I am always or 

almost always sometimes sometimes almost always 

satisfied satisfied dissatisfied dissatisfied 

with. . . with. . . with. . . with. . . 



(Dav and Bodur 1978; 1979; 



10. I am quite I am somewhat I am somewhat I am quite 
satisfied satisfied dissatisfied dissatisfied 
with... with... with... with... 



(Day and Bodur 1979) 



Figure 4 (a) --continued 



51 



11. Completely Very Somewhat 

Satisfied Satisfied Satisfied Satisfied 

(Dissatisfied) (Dissatisfied) (Dissatisfied) (Dissatisfied) 
12 3 4 



Not 
Satisfied 
(Dissatisfied) 
5 



(Aiello & Czepiel 1979; Aiello, Czepiel and Rosenberg 1977) 



12. Very Somewhat Neither Satisfied Somewhat Very 

Satisfied Satisfied Nor Dissatisfied Dissatisfied Dissatisfied 
12 3 4 5 

(Aiello, Czepiel and Rosenberg 1977; Mowen and Grove 1983) 



13. Now that you've actually used the product, how satisfied with it 
are you? 

Dissatisfied Satisfied 

" (Bahr 1982) 



14. Completely Fairly Not too 

Satisfied Satisfied Satisfied 

(Hughes 1977. 



5. I am satisfied with . 

Agree _ _ : _ : : _ _ • _ : _ • _ _ : • _ _ Disagree 

(Oliver and Bearden 1983) 



Other evaluations 

16. To what extent does this __ meet your needs at this time? 

Extremely Extremely 

Well __:__:__:_ : : : __•'__ : Poorly 

(Oliver and Westbrook 1982; Westbrook 1980b) 
Figure 4(a)--continued 



17. Summed semantic differential scales of satisfaction. 

(Oliver and Bearden 1983; Oliver and Westbrook 1982; 

Westbrook and Oliver 1981) 



52 



18. Likert Scales 



(Oliver 1980a) 



19. Satisfactory or Dissatisfactory occasions/products as judged by 
respondent. 

(Day and Bodur 1978; Locker and Dunt 1978; Richins 1983a) 



Graphic 

20 



Imagine that the following circles represent the satisfaction of 

different people with _ . Circle has all minuses in it, to 

represent a person who is completely dissatisfied with . 

Circle 8 has all pluses in it, to represent a person who is 
completely satisfied with . Other circles are in between. 




Which circle do you think comes closest to matching your 
satisfaction with ? Write the circle number here: 



(Oliver and Bearden 1983; Oliver and Westbrook 1982: 
Westbrook 1983; Westbrook and Oliver 1981) 



21. Here is a picture of a ladder. At the bottom of the ladder is the 
worst _ you might reasonably expect to have. At the top is the 
best _ _ you might expect to have. On which rung would you put 



9 Best I could expect to have 

8 

7 

6 

5 

4 

3 

2 

1 Worst I could expect to have 

(Andrews and Withey 1976) 



Figure 4(a) — continued 



53 



Verbal 

22. Likert scales 

a. I am satisfied with 



b. If I had it to do all over again, I would 

c. My choice to was a good one. 

d. I feel bad about my decision concerning 



I think that I did the right thing when I decided . 

f. I am not happy that I did what I did about . 

Agree ...(9) ... (7) ... (5) ... Disagree 

Strongly Agree Strongly Disagree 

(Moore & Shuptrine 1984; Oliver 1980a; Oliver and Bearden 1983; 
Oliver and Westbrook 1982; Westbrook and Oliver 1981) 



23. Mark on one of the nine blanks below the position which most 
closely reflects your satisfaction with 



Delighted Pleased Mostly Mixed Mostly 
Satisfied Dissatisfied 



Unhappy Terrible Neutral Never Thought 

About It 
(Jordan and Leigh 1984 [7 items] ; Moore and Shuptrine 1984: 
Oliver and Bearden 1983; Oliver and Westbrook 1982: 

Westbrook and Oliver 1981] 



24. Content analysis of subject-provided protocols with scoring for 
satisfaction or emotional statements. 

(Locker and Dunt 1978; Westbrook 1980b) 



25. Scales measuring separate emotions. 

Please indicate the extent to which each word describes the way 
you feel with respect to 

12 3 4 5 

Very Slightly Slightly Moderately Considerably Very 

or not at all Strongly 



(b) Emotional/Affective Measures in Consumer Satisfaction 



54 



Adjectives "loading" on each of the ten emotional dimensions d 

1) Interest - Excitement, 2) Enjoyment - Joy, 3) Surprise - 
Startle, 4) Sadness - Anguish, 5) Anger - Rage, 6) Disgusted - 
Revulsion, 7) Contempt - Scorn, 8) Fear - Terror, 9) Shame - 
Shyness and 10) Guilt - Remorse 

(Westbrook 1987; Westbrook and Oliver 1984) 



Graphic 

26. How do you feel about 
I feel: 

7 



3 



Delighted Pleased Mostly Mixed Mostly Unhappy Terrible 

Satisfied (about Dissatisfied 

equally 

satisfied 

and 

dissatisfied 



A Neutral (neither satisfied nor dissatisfied) 

I never thought about it. 

(Andrews and Withey 1976; Westbrook 1980b) 



Figure 4 (b)--continued 



55 



27. "Feeling" Thermometer 

Where would you put on the feeling thermometer? 

WARM 100° — Very warm or favorable feeling 

85° — Good warm or favorable feeling 

70° — Fairly warm or favorable feeling 

60° -- A bit more warm or favorable than cold 
feeling 

50° -- No feeling at all 

40° -- A bit more cold or unfavorable feeling 
30° -- Fairly cold or unfavorable feeling 

15° -- Quite cold or unfavorable feeling 



COLD 



0o — Very cold or unfavorable feeling 
(Andrews and Withey 1976; Oliver and Westbrook 1982; 

Westbrook and Oliver 1981) 



28. Faces scale 

Here are some faces expressing various feelings. Below each is a 
letter. 





A B C D E F 

Which face comes closest to expressing how you feel about 



(Churchill and Surprenant 1982; Oliver and Westbrook 1982; 

Westbrook and Oliver 1981) 



Figure 4 (b) --continued 



56 



Verbal 

Behavioral intentions 

29. Because of _ I would come (shop) here again. 

Very Strong Strong Strong Very Strong 

Yes Yes Yes ? No No No 

(Swan7 _ Trawick~and Carroll 1981) 



30. How likely are you to play with (use) _ in the future? 

Very Very 

Unlikely Unlikely Likely Likely 

-2 -1 +1 +2 



(Jordan and Leigh 1984] 



31. Knowing what you know now, what are the chances in ten (10) that 
you would choose to use the again? 

0123456 789 10 
No Chance Certain 

(Oliver and Bearden 1983; Westbrook and Oliver 1 98 1 ) 

Graphic (Observational) 

32. Measures of time and extent of use. 

(Bjorklund and Bjorklund 1979) 

33. Filing complaint as sign of dissatisfaction. 

(TARP 1979) 

34. Loyalty, repurchase as sign of satisfaction. 

(LaBarbera and Mazursky 1983) 

Notes to accompany Figure 4: 

a Citations provided are meant to serve as examples of scale use, 

not to provide an exhaustive bibliography. 
b Some parenthesized modifications of what was essentially the same 

scale are presented for the sake of simplicity. 
c In this case, the scale was compressed by omitting the neutral 

responses . 
d Again, for the sake of simplicity, all cf the adjectives which 

were used are not presented in this Figure. 

(c) Behavioral/Conative Measures in Consumer Satisfaction 



57 



depicted in both survey and experimental research. As such, it is 
important to note several issues which arise in their use: (1) Each 
requires an interpretation by the respondent and questioner of the word 
"satisfaction" or some variant thereof. As noted in the beginning of 
this chapter, the word has a variety of meanings even in academic 
research. (2) Where a midpoint is included, it is often not marked 
(e.g., measure #13), marked in an ambiguous manner (e.g., measure #6), 
or marked in apparent conflict with other measures (e.g., measure #5 
hypothesizes a mixed evaluation whereas measure #7 suggests a lack cf 
evaluation). (3) Not providing for a neutral or mixed response (e.g., 
measure #9) forces a possibly spurious response. (4) The range of 
gradations provided, from dichotomous (e.g., measure #8) to a 13-point 
scale (e.g., measure #13) suggests different evaluation processes. 

The measures classed as "other" provide a slightly different 
perspective. (1) The need scale (measure #16) is somewhat more 
evaluative than the expectation disconf irmation scales in the first 
group. (2) Although the specific adjectives used for the semantic 
differentials (measure #17) were not provided by the original authors, 
related work by Westbrook and by Oliver suggests that this particular 
technique may alleviate some of the definitional ambiguity mentioned 
above by the use of synonyms in the measurement of aspects of 
"satisfaction". (3) The last type of "measurement" in this group 
encompasses those open-ended survey techniques that ask the respondent 
to first identify a satisfactory or dissatisfactory critical incident 
and then provide more details about it. 

The graphic measures of satisfaction as an evaluation, presented 



58 



at the end of Figure 4a, probe for essentially the same information as 
the verbal measures. The graphic format, however, is better able to 
communicate the concept of quantities of satisfaction 
(dissatisfaction), thus removing some of the ambiguity of the verbal 
measures. This technique also makes the distinction between mixed 
evaluations and the absence of evaluation more evident. 

Figure 4b presents measures that are more consistent with the 
emotional definition of satisfaction that has been adopted for the 
dissertation. The Likert-type scales (measure #22) were defined to be 
"emotional in content" when presented by Oliver (1980a, p. 463) despite 
some resemblance to the evaluative scales in part (a) of the Figure. 
The next scale (measure #23) was derived from Andrews and Withey's 
(1976) set of indicators of well-being (as were a number of the 
measures being reviewed) . This single scale seems to capture both the 
evaluative and emotional aspects of satisfaction in its labels and 
seems to provide ample options for neutral responses (although this is 
sometimes defeated by omission of the off-scale response categories; 
e.g., Jordan and Leigh 1984). It is not clear from the CS/D 
literature, however, whether or not the scale possesses more than 
ordinal properties (e.g., interval which has been presumed for some of 
the other scales) . 

A recent innovation in measurement within the CS/D literature is 
the technique being developed by Westbrook (1983) to assess individual 
emotions at times or occasions that are critical in the satisfaction 
process. Although still in preliminary stages, these measures hold 
promise as more direct, but less obtrusive measures of satisfaction 



59 



than the evaluative scales presented previously. Figure 4b shows only 
a portion of the measurement scheme (measure #25) that has been 
adapted from Izard's (1972) Differential Emotions Scale. 

The original scale utilized responses to 69 separate items 
describing a person's emotional profile at a particular time. 
Responses to specific items are summed and the resultant profile is a 
subset of the ten emotional dimensions listed in the Figure. Westbrook 
has used a truncated, 31-item version of the scale to describe 
consumers' emotions with respect to their automobiles. Factor analyses 
have revealed four (westbrook 1983) to six (Westbrook and Oliver 1984) 
distinct profiles of emotional responses. How these findings relate to 
the emotional theory developed in the dissertation is explored in the 
beginning of Chapter 3. 

Measurement of emotions is difficult because emotions are 
difficult to verbalize and degrees of emotion are hard to represent. 
Thus, the graphic emotion instruments (measures #26-28) offer some 
simplification in representing a gestalt reaction. 

Finally, satisfaction has been defined as the precursor to 
specific behaviors (Cardozo 1965) and is being treated in the 
dissertation as theoretically linked to such behaviors (Figure 3). 
Figure 4c outlines behavioral measures that have been used as measures 
of satisfaction. Even if these are not justified as direct measures of 
feeling, the causal flow that has been proposed suggests that behavior 
may be indicative of strong feeling. Note that in the case of Figure 
4c, the graphic measures are observations of actual behavior (measures 
#32, #33).. This categorization preserves the composite, less dependent 



60 



on connotation, characteristics of previous graphic measures. 
Review and Usefulness of Measures 

There have been prior efforts at reviewing and empirically 
comparing satisfaction measures in the CS/D literature. Hunt (1977a;b) 
commented on the discussions at the first two CS/D conferences and 
identified the need for a common measurement methodology as a key to 
deriving integrated CS/D theories. He seemed to favor conative verbal 
measures as "a composite measure getting at all the influences 
affecting the decision without having to identify those influences 
. . . If a repeat purchase is intended, that says that all things 
considered, the purchase was critical enough that it has to be repeated 
and the choice was good enough that the respondent doesn't think any 
better choice is worth making." (1977b, p. 39). Note that the use of 
behavioral intentions rather than actual behaviors as measures reduces 
concerns about influences external to the satisfaction situation that 
could modify behavior (Fishbein 1966; Fishbein and Ajzen 1975). 

Westbrook and Oliver (1981; Oliver and Westbrook 1982) have 
compared a number of the measures depicted in Figure 4 empirically (the 
dimensions in the Figure are based, in part, on discriminations by 
these authors). From two analyses of pilot data they conclude (1) 
Likert, semantic differential and a composite verbal scale 
(corresponding to measures #22, #17 and #5 + #20b + #23 from Figure 4, 
respectively) performed best on convergence versus divergence criteria; 
(2) discriminability of various scales seemed to be product class 
dependent (automobiles versus calculators); and (3) as a whole, the 
measures did not succeed very well in discriminating satisfaction from 



61 

attitude. The first finding is encouraging for those who wish to 
identify measures which will be generally useful. The latter two 
findings threaten the universality of both the measures and theories of 
satisfaction. 

The conclusions by Westbrook and Oliver are tempered greatly by 
their methodology and the ensuing discussion. Assuming that a student 
sample (n=160) is sufficiently representative for this type of research 
(theory testing), the manner of questioning still leaves room for 
criticism. Essentially, the students were asked to give their present 
reaction to durables which they may have possessed for some time and 
with which they are likely to have varying histories of ownership and 
usage experience. It is not surprising, then, that the measures 
reflect some difference in response to product class and some 
confounding with attitude. The authors allude to an explanation 
"satisfaction . . . is, in effect, a response to disconf irmat ion . . . 
and is expressed in affective terms .... In a sense, satisfaction 
may be seen as a disturbance acting on an attitude system" (1982, 
p. 13). They go on to suggest that temporally distinct measures may 
help to separate satisfaction and attitude. Given Oliver's (1981) 
argument that satisfaction is likely to rapidlv "decay" into attitude. 
it would seem that temporally distinct methodology would be necessary 
to uniquelv identifv the constructs. 

In the absence of such temporal distinction, the cautionary 
conclusions by Westbrook and Oliver are suspect. This inattention to 
the theory relating satisfaction and attitude is continued in the work 
by the same authors in which they attempt to calibrate the emotional 



62 



measures (measure #25) derived from the Differential Emotions Scale. 
In this research (Westbrook 1983; Westbrook and Oliver 1984) they 
repeat the methodology comprised of collecting contemporaneous 
reactions to products (i.e. automobiles) which have been owned and used 
over a range of time periods, and usage patterns. The resulting 
emotional dimensions are correlated with more typical satisfaction 
measures (e.g., measures #7, #19, #20, #33) with mixed results. Again, 
the theoretical distinction between satisfaction and attitude is not 
maintained and the influence of time is disregarded. 6 

Similar concerns about timing versus attitude formation have also 
been expressed outside the "mainstream' 1 CS/D literature (Locker and 
Dunt 1978). In this review of British studies concerning the 
measurement of patient satisfaction with medical care, the authors 
identify many of the same techniques that are listed in Figure 4. They 
also are concerned about the necessity of developing universal measures 
of satisfaction to allow for inter-study comparability. The interview- 
technique seems to be much more prevalent in this literature, and the 
findings suggest that global evaluations exposed by direct measurement 
scales seriously mask the separate underlying attribute evaluations. 

This argument returns us to the levels-of-specif icity question 
that was discussed previously. In order to be useful in theory 
testing, measures must be applicable at the same level of specificity 
as the operational theory. Locker and Dunt (1978) argue for examining 



6 The measurement technique proposed here is used in conjunct ior 
with the critical incident methodology, measures #19 a. id #24, in an 
attempt to alleviate this time problem in the derivation of measures 
for the dissertation. This is discussed in more detail as the Pilot 
Study in Chapter 5. 



63 



various aspects of the patient's experience with medical care in order 
to explain his satisfaction, much the same as Wanous and Lawler (1972) 
examine the influence of various aspects of a job on job satisfaction. 
Although these are likely to be useful for understanding which factors 
influence satisfaction (maybe in a specific situation), it is less 
likely that a respondent will be able to discriminate the separate 
influences on an inventory of emotions such as that presented by 
Westbrook (1983) in order to build a "net" satisfaction. The final 
result of this may be that the emotional measures (Figure 4b) will best 
reveal the presence and degree of satisfaction whereas the evaluative 
measures (Figure 4a) will be more useful for explaining the cause of 
the feeling (i.e. it may be difficult to make emotional measures 
salient at more than one level in a given measurement situation). 

Even so, these measures should not be used without additional 
cautions. Miller (1977) argued that attempts to measure a process may 
actually impact on the process being studied. This problem of reactive 
measures and concomitant demand artifacts has been discussed elsewhere 
in consumer behavior (Sawver 1975) but the arguments do not seem to 
have effected the design of less obtrusive measures (as suggested by 
Webb, Campbell, Schwartz, and Sechrest 1956 and Sechrest 1979). Sobei 
and McGuire (1977) have gone so far as to argue that disconf irmation 
(as the difference between expectations and perceptions--discussed here 
in a slightly different context) mav be simply a measurement artifaci. 

The possibility of this, given current operat ionalizat ions of the 
satisfaction theory that has been presented, should not be overly 
discounted. Some authors have reported the necessity of forcing a 



64 



satisfaction response by eliminating the opportunity for neutrality 
(e.g. Day and Bodur 1979; Jordan and Leigh 1984). Although this may be 
viewed as forcing more careful responses, it is argued in Chapter 3 
that such a procedure obscures important theoretical differences 
between a mild response and no response. For our purposes, it is 
important to note that the treatment of neutral options such as "Don't 
Know" or "Never Thought About It" has been shown to bias recall data 
(Mizerski, Freiden and Greene 1983) and can be presumed to have similar 
effects on attitudinal or satisfaction responses. 

For these reasons, observational measures such as those suggested 
in Figure 4c may actually be better for some purposes than the more 
direct satisfaction measures identified elsewhere in the Figure. 
Although these behavioral measures may be influenced by other externa] 
factors, the causal flow- may be sufficiently strong to drive 
discernible, theoretically important differences. The influence of 
satisfaction on such behaviors as' word-of-mouth, complaining, and 
repurchase is examined in detail in these separate literatures. The 
next section of this chapter describes how the concepts central to 
these literatures relate to CS/D as the products of the emotions. 
Behavior al Outcomes of Satisfaction 

As discussed previously, it was the observation of certain 
marketplace behaviors that first aroused interest in satisfaction 
(Cardozo 1965) and has continued to fuel the growth and development of 
the CS/D literature. In this section, the theories and research that 
examine these behaviors and implicate the satisfaction feeling as a 
cause are introduced. Further, it is suggested that the continuum of 



65 
behaviors proposed is sufficiently stable within the CS/D literature to 
be useful as a gross measure of relative satisfaction. 

Day (1977) presented the most integrated discussion of behaviors 
thought to follow feelings of satisfaction. Although he presented 
separate taxonomies for satisfaction responses and dissatisfaction 
responses, the distinctions made were the same and it is more in 
keeping with the unidimensional nature of the construct to summarize 
his taxonomy in a single hierarchy (Figure 5). Day's presentation of 
the hierarchy in a form reminiscent of a decision tree is an apt 
metaphor for describing the research that examines these behaviors. 
That is, the literature tends to view these behaviors as resulting from 
rational choice processes among economically evaluated alternatives. 

The behaviors are generally treated as spanning a continuum which 
seems to suggest that, all else equal, a stronger satisfaction feeling 
(positive or negative) motivates a stronger response that may require 
more effort or other forms of "cost" from the individual (Figure 6). 
For example, in most cases it is less convenient for a consumer to 
contact either the marketer or a third-party complaint-handler than it 
is to express his/her dissatisfaction to a friend. This Figure also 
expresses a concept that is frequently overlooked in the CS/D 
literature; responses to satisfaction are analogous to the responses to 
dissatisfaction (negative satisfaction). The latter behaviors 
(especially complaining) are more visible and more troublesome to 
marketers and have garnered the majority of the research interest. 
Yet, consumers definitely do exhibit overt positive responses to 
feelings of satisfaction. Day (1977) reported that many companies 



66 






01 




01 U 




0> 01 c 




, > 01 S H 




<H 3 4J 




O B*< 




in o tiO 




D 4J 4J 




05 i-i 



HI 




V, 




TJ 




TJ 




c 


a) 


C 


M 


aj 


a 


id 


01 


•H 


jj 




iH 


M 




w 




h 


jj 


u 


11 




3 


3 


en 


c 


TJ 




M .Q 


() 


M 


id 


< 


Ul 





2 




a, 





a-u 





i-i 





U 


4-> 


o> 


4-1 


■3 






W TJ 


M 


^H 










oj 


O 


M 




ca 


4-1 


Dh T) 








G 


4J 


0) 


01 fl 


4J 


T) 


c 


U 


O 


■H 


•H 


DO 


u 


[,J 


>- 




Sn 


01 


3 


i-i 


O 


Q 


aa 





03 







i 


01 






c 


OJ 




■H 


Li 


■H 




1-1 


01 


t) 


O 


CU 


> 


c 


u 







QJ 






o 


tn 


4 C 


0] 




< 


1 '" 1 


m 


1-1 




fl 


01 







H 


c 




fl 


a 


•H 


0) 


4J 


h 


01 


AJ 


C 





3 


fl 


01 


u 


in 


> 


S 







01 






01 


-4 




01 


* 


n 


1-1 


014-1 TJ 


01 




01 


J 


c 




LB 


01 


~4 


c 


.X 


4-1 


— 1 


fl 





fl 


H 


< 


4J 

a 







tn 


01 

01 


S 




B 


01 


i-i 


•H 


1-1 


bu 


lib 


T) 






01 


>i w 


as 




01 




4J 


01 


.* 


u 


c 


0> 


0) 


■H 


0) 


I-I 


01 


01 


■H 


3 




a 


CO 



e 

0) 01 

01 01 4-1 

<y 3 m 




3 

£ ,3 



Max imum Max imum 

Dissatisfaction Satisfaction 



Complaining Negative Exit Loyalty Positive Compliment 
(seek Word-of- (Dis- (Degree Word-of- 
redress) Mouth position) of Use) Mouth 



Figure 6 
Satisfaction Responses 



6' 



63 



that operate "consumer hot lines" routinely receive complimentary 
calls about their products and services. 

Extreme actions are comparatively uncommon, however. In this 
section, the analysis of the literature on complimenting and 
complaining follows discussions of loyalty and exit, and positive and 
negative word-of -mouth. Some articles have examined relatively 
abnormal behavior such as destruction of property and merchandise (Mil is 
1981), certain forms of consumer fraud (Wilkes 1978), as well as 
shoplifting (Herndon 1972; Klemke 1982; Moschis, Cox and Kellaris 
1987). While these behaviors have been characterized as retribution 
for dissatisfaction and may result from the process under discussion, 
they are not sufficiently common or well-understood to be relevant for 
this discussion. 

The behaviors which are most likely to be affected by weak 
feelings of satisfaction are identified in the center of the continuum 
(Figure 6) as Loyalty and Exit. The specific behaviors that are 
encompassed by these classifications (e.g. repeat purchase) tend to 
arise naturally as part of a consumer's experience. Satisfaction is 
thought to influence the choice of behavior, perhaps through an 
attitudinal process, rather than to initiate the performance of 
specific acts. 

The consumer decisions that might be influenced at the Loyalty 
versus Exit level can be categorized under three classifications: (1) 
Use versus Do Not Use; (2) Retain versus Dispose; and (3) Repurchase 
versus Do Not Repurchase. Choices among these classes of behavior are 
generally thought to be determined by attitude which, in turn, has been 



69 



argued to be related to satisfaction (Oliver 1 98 1 — the nature of this 
relationship is explicated more fully in Chapter 3). 

Usage decisions have been related to satisfaction generally in the 
context of repeatedly used services such as electrical utilities 
(Heberlein, Linz and Ortiz 1982), public radio station broadcasts (Swan 
and Trawick 1983) and leisure activities (Hawes, Blackwell and Talarzyk 
1976). Arguably, such consumption actually entails a series of 
repeated "purchase" decisions that are affected more by attitude than 
by satisfaction. These papers serve to introduce the argument that a 
feeling of satisfaction on one usage occasion may prompt additional 
usage whereas a feeling of dissatisfaction may reduce usage. For 
example, the extent of use of a new home computer may be related to the 
satisfaction feelings generated by the first few trials. 

Similarly, the decision to retain or dispose of a product can 
hinge on one's attitude toward the item and on the feelings of 
satisfaction which may have contributed to that attitude (or, even, 
dissonance after purchase but before delivery-- Donnelly and Ivancevich 
1970). DeBell and Dardis (1979) conveyed similar thoughts in their 
argument that product failure may hasten the perception of obsolescence 
and disposal of major appliances despite the possibility of or actual 
repair of the item (also, Burke, Conn and Lutz 1978a,b). Many of the 
items placed for disposal are still serviceable as evidenced by the 
popularity of flea markets ("swap meets" and other terms are also used 
--Belk, Sherry and Wallendorf 1988; Wallendorf and Arnould 1988). 
Unfortunately, there is little other research concerning disposal 
decisions on which to base further argument. 



70 



In the consumer context, the term "loyalty" most often suggests 
some form of repeat purchase behavior and is generally associated with 
the purchase of a particular brand in a particular product form 
(Bernardo and Tarpey 1980; Goering 1985). The literature on brand 
loyalty is copious (see Jacoby and Chestnut 1978 for a review) and is 
not detailed here. It is useful to note that consumers' prior 
satisfaction experiences have been linked explicitly to repurchase 
(Oliver 1980a; Oliver and Linda 1981) and current evaluation (Cohen and 
Houston 1972). This can result even when the target of the 
satisfaction feeling and the subsequent purchase are related, but not 
identical, as seen with choice of housing type (Leigh 1984), warranty 
service and automobile purchase (Bernacchi, Kono and Willette 1980), as 
well as the purchase of different small appliances from the same 
manufacturer (Smead, Wilcox and Wilkes 1980). 

Direct repeat purchase decisions have been described for diverse 
products from automobile repair (Biehal 1983) to children's videogames 
(Jordan and Leigh 1984). Schmalensee (1978.) suggested a phenomenon 
that is essentially identical to the disconf irmation paradigm when he 
proposed a method of modeling advertising effects on repeat purchase. 
The upshot of this, then, is that experience with prior purchases is 
likely to influence future purchases, whether of identical or merely 
related items. LaBarbera and Mazursky (1933) for example, reported 
that much of the decision making for commonly purchased products was 
completed before the actual shopping situation (see also Hover 1984). 



The satisfaction feelings that result from repeated use of such 
products can act to maintain or adjust the consumer's attitude which, 
in turn, influences the next purchase decision. 

Finally, it should again be noted that although the preceding 
discussion has centered on discrete products and brand loyalty, there 
is no reason to suspect that the effects of satisfaction are so 
limited. As discussed previously, there are a number of levels of 
specificity at which the target of the satisfaction feeling may be 
identified. Given the appropriate opportunity, the consumer may 
respond by remaining loyal (repurchasing, maintaining a business 
relationship) or by exiting (switching to another supplier, product 
form, etc.). Carey, Clicque, Leighton and Milton (1976), for example, 
suggested that managing customer satisfaction may be an effective means 
of garnering repurchase for a retailer. 

Whereas the opportunity to select between Loyalty and Exit 
behavior arises in the conduct of the consumer's regular life, as 
mentioned in the preceding discussion, other responses to satisfaction 
r eelings require the consumer to initiate specific actions. The next 
class of behaviors represents an increase in the specific motivation 
required. These behaviors have been termed Word- of -Mouth, and again, 
can entail either positive or negative reactions. 

Marketers first became interested in the positive word-of-mouth 
which was identified as helpful in enhancing the diffusion of 
innovative products (Arndt 1967; Engel, Kegerreis and Blackwell 1969; 
Sheth 1971). This diffusion literature eventually yielded the concept 
of opinion leaders (individual consumers who heavily influenced the 



"2 
attitudes and behaviors of their associates) that eventually became 
the more general concept of social influence within reference groups 
(Burnkrant and Cousineau 1975; Moschis 1976; Park and Lessig 1977). 
The conclusion from this research is that consumers exchange 
information about consumption experiences among themselves and that the 
subsequent behaviors of the recipients of such information show an 
influence attributable to the communication. 7 Moreover, it has been 
suggested that evaluative information about the product is more 
influential than signals of social acceptability or group preference 
(Burnkrant and Cousineau 1975). This suggests that the product 
expectations and perceptions central to the disconf irmation paradigm 
are related to communications among consumers (Coca-Cola 1982). 

Researchers in CS/D eventually became more interested in 
investigating the causes cf word-of-mouth behavior rather than its 
effects on the recipients of the communications. Day's (1977) taxonomy 
(Figure 5) categorizes word-of-mouth behavior as "private action" 
although it may be viewed as considerably more public than one's own 
internal resolve to increase or decrease purchases or use cf an item 
(except in cases where use constitutes a form of word-of- 
mouth— -Koepp 1985). No matter how this particular distinction is 
ultimately resolved, the placement on the continuum of responses 
(Figure 6) is quite consistent with the treatment of word-of-mouth in 
CS/D (e.g. Lewis 1983). Richins (1983a, b) is more specifically 
interested in negative word-of-mouth as a response to dissatisfaction. 



7 Westbrook (1987) treats W-O-M and public opinion rumors as 
essentially similar. The emotional treatment of satisfaction/ 
dissatisfaction suggests different motivations, however. 



73 



She reported that (1) the likelihood of any response increases with 
degree of dissatisfaction, and (2) word-of-mouth tends to be more 
common than complaining to the seller. Furthermore, behaviors 
categorized as word of mouth can range from a simple mention to an 
extensive, animated discourse. The kev criterion is the consumer's 
motivation to act this way. 

Of course, word-of-mouth behaviors are not solely dependent on the 
valence and extent of the satisfaction feeling but are also influenced 
by other aspects of the situation. Research attempting to describe the 
role of these other determinants has focused on the extremes of the 
continuum presented in Figure 6, Compliments and Complaining (sometimes 
called Voice in other satisfaction literatures). 

In contrast to the classes of behaviors just discussed, the 
literature concerned with this last class of behaviors has originated 
with and concentrated on negative rather than positive behaviors. 
Although ccmpliments arise in measurable amounts (Day 1977; Lewis 1983) 
it is likely that they are viewed as financially or economically less 
useful to the firm than loyalty or positive word-of-mouth. On the 
other hand, complaints imply a greater extent of switching and negative 
word-of-mouth and generally include requests for redress which require 
immediate attention. Thus, it is not too surprising that there is a 
voluminous literature on complaining behavior itself. 

This literature addresses such diverse topics as the extent of 
complaining (TARP 1979), the influence of complaints on business 
practices (Cobb, Walgreen and Hollowed 1987; Fornell and Wernerfelt 
1987; Resnik and Harmon 1983; Ross and Gardner 1985) and the subsequent 



effects cf business response on the attitudes and behaviors of the 
complainants (Bearden and Oliver 1985; Gilly and Gelb 1982; Wilkinson 
and Mason 1979). At issue in this review however, is the research 
that seeks to identify the causes of complaining or, more correctly, 
the selection of complaining as a response to dissatisfaction (Figure 
5). 

Consumer complaining behavior is not a recent phenomenon; for 
example, legend has it that the young Michelangelo once counterfeited a 
statue and that, on learning of the deception, the buyer returned the 
item and demanded a refund (MacDougall 1958). The necessity of 
avoiding such negative impact on revenues has fueled a great deal of 
interest among marketers in complaining behavior. 3 This type of 
behavior can become even more extreme and more costly if reparations 
are sought via the legal system (Best 1981; Brown and Swartz 1984; 
Goodwin, Mohajan and Bhatt 1979; Morgan and Avrunin 1982). 

A number of attempts have been made to identify causes or factors 
leading to complaining. In the typical study, groups of dissatisfied 
consumers are divided into those who did complain and those who did net 
(e.g., Bernacchi, Kono and ivillette 1980). The groups have been 
compared on such dimensions as demographics — age, education, sex, 
income, etc . --(Bearden and Mason 1984; Morganosky and Buckley 1987), 
personality (Fornell and Westbrook 1979), prior experience with 
complaining (Gronhaug and Zaltman 1981) and a belief-based attitude 



b "Complaining" will generally include returns for repair, 
replacement or refund, communicating displeasure to the seller or other- 
channel member without a specific request or desire for redress; and 
similar communications addressed to some "neutral" third party such as 
government agencies or newspaper "Action Lines"--Feick 1984. 



75 



model that implicates specific beliefs about the consequences of 
complaining (Richins 1982). 

The literature has not provided firm conclusions concerning the 
influences of such variables (Bearden 1983; Oliver 1987); the results 
are mixed for each. Some problems have been definitional. For 
example, there is little to distinguish a personality variable labelled 
"propensity to complain" from a cognitive variable labelled 
"complaining experience" when each is defined by past complaint 
behavior (Robinson, Trebbi and Adler 1983). Moreover, it is difficult 
to account for the effects of the degree of dissatisfaction, which is 
sometimes significant (Zaltman. Srivastava and Deshpande 1978) arid 
sometimes not significant (Gilly and Gelb 1982) in determining 
complaint behavior (see also Oliver 1987). 

Another possible reason for these inconsistencies is the failure 
to view complaining as one element on a continuum of possible responses 
(Figure 6). Although not all authors have treated the choice to 
complain as dichotomous (Jacoby and Jaccard 1981), some have (Fornell 
and Westbrook 1979, 1934; Gronhaug 1977) which may be contributing to 
the discrepant findings in the literature (Richins 1987; Singh 1988). 

Iable 3 is a somewhat more complete list of alternative behaviors 
than that provided by Day's taxonomy (Figure 5). The list was compiled 
from surveys that were concerned with consumers' post dissatisfaction 
responses. Although there has been interest in compliments and other 
positive behavior (Robinson and Berl 1 980), a similar inventory of 
responses to satisfaction has not been reported. Treatment of the 



76 



Table 3 
Actions Following Dissatisfaction 



Consumer Action 



Stopped payment or refused to pay 

Decided not to buy the product or service or deal with that compam 
again 

Complained to family or friends 

Asked for replacement or refund 

Complained to the person who sold me the product or service 

Complained to the company or store 

Complained to a consumer agency 

Complained to a public agency or my congressman 

Complained in a letter to a newspaper or magazine (for publicity 
rather than redress) 

Considered taking legal action 

Consulted or hired a lawyer to protect my interests 

Took no action at all 

Note: Table based on Dav 1983 and Krishnan and Valle 1979. 



77 



relationships among possible behaviors and the similarities of positive 
and negative responses has recently been integrated into models that 
encompass both the satisfaction and complaining processes (Bearden and 
Teel 1983; Day 1977; Lippert and MacDonald 1981). 

The result of the integration of these theories is a view of post- 
evaluative behavior that posits that the degree of satisfaction 
emotion which is felt determines the level of drive or motivation to 
respond; the relative costs and benefits that are relevant for the 
particular individual in the particular situation determine the choice 
of response (Day 1984; Feick 1987; Fornell and Didow 1980; Richins 
1932). Thus the factors identified previously (demographics, 
personality", etc.) can be reinterpreted as affecting choice of behavior 
by first affecting the expectation of costs or benefits (e.g. prior 
experience may increase the expected gain from complaining in certain 
situations and decrease the cost of learning how, where and to whom to 
complain). Consumer handbooks designed to improve the public's ability 
to complain typically capitalize on just such an economic concept by- 
illustrating how much can be "won" through complaining (Charell 1973: 
Horowitz 1979) and/or bv making complaint handlers more accessible 
(Eisenberger '977; Feinman 1979; Knauer 1983). 

Finally, this economic approach also accounts for the discrepancy 
in levels of positive versus negative behaviors. Although some measure 
of the degree of satisfaction may indicate equivalent levels, the 
cost:benefit ratio of complaining in a given situation is not likely to 
equal that of conveying a compliment, for example. The continuum 



presented in Figure 6 incorporates this cost versus benefit concept at 
a very gross level. 

It is the assumption that this behavioral continuum holds across 
situations that underlies the usefulness of complaint data as a 
surrogate for measuring dissatisfaction. Comparisons across products 
and individuals have shown that this is inappropriate. In a particular 
situation, with better defined or constant costs and benefits (such as 
a laboratory experiment) the use of behavior as a measure of 
satisfaction may be more justified. The loss of precision may, in many 
cases, be balanced by the gain in the unobtrusiveness of collecting 
behavioral (versus rating scale) data. 

S ummary 

This chapter has defined the satisfaction construct as an 
emotional response and adopted one model of the satisfaction process 
from the CS/D literature. The types of consumer satisfaction that have 
been addressed by researchers v:ere shewn to be specific manifestations 
of the general construct and model, made specific by identifying the 
target of the emotions. The advantages and disadvantages of various 
measurement techniques were discussed, especially their conformity to 
the definition and model. Finally, behaviors which result from 
satisfaction were identified and the question of their feasibility as 
measures was raised. 

The next chapter extends this satisfaction construct, as defined 
and modeled here, bevond the consumer behavior framework. It is argued 
that a basic human behavioral process results in a basic feeling of 
satisfaction, which has been observed in consumer situations. The 



79 
present framework is extended to identify other factors that are 
necessary to evoke the emotion. This extended conceptualization is 
then specifically applied to consumer behavior and the CS/D literature. 



CHAPTER 3 
SATISFACTION: AN EXTENDED CONCEPTUALIZATION 
OF THE GENERAL CONSTRUCT 

Introduction 

In this chapter, the CS/D literature is augmented with theories 
and research findings from other fields. The concepts that are 
borrowed clarify the nature of the pattern of emotions which defines 
satisfaction and which can be aroused in a variety of contexts, only a 
subset of which involve behavior as a consumer. It is further argued 
that satisfaction is not aroused by all behavior. As with attitudes, 
satisfaction with a given target frequentlv may not exist for a 
particular individual. As an emotional construct, satisfaction is 
thought to dissipate relatively rapidly but not without affecting 
attitude toward the target object. This relationship is explored and 
implications for the time-sensitivity of satisfaction measurement are 
discussed. 

Given the uncertainty regarding arousal of the satisfaction 
feeling and its fleeting nature once present, it is necessary to 
understand and be able to predict its occurrence. The major 
contribution of this chapter, and the dissertation as a whole, is the 
presentation cf six necessary and collectively sufficient conditions 
for the arousal of the general feeling of satisfaction. The 
explication of these antecedent conditions is followed by a discussion 
of consumer situations in which they are likely to be present and an 

30 



examination of CS/D research in which the likelihood that the 
satisfaction feeling was aroused in a given study is assessed in light 
of the antecedent conditions and the measures used for satisfaction. 
The remainder of the dissertation examines two of the antecedent 
conditions empirically. 

The philosophy that has been adopted views satisfaction as one 
construct (MacCorquodale and Meehl 1948) in an integrated network. The 
theory that is developed concerns the general satisfaction construct. 
The empirical portion of the dissertation addresses the more specific 
consumer oriented construct and theory, derived from these theories (S. 
Hunt 1976). It is crucial that the distinction be maintained between 
the term "satisfaction," signifying the general emotional construct, 
and the labels "CS/D" or "consumer satisfaction", signifying a 
literature that applies the theory to a- range of consumer behaviors. 
Definition and Theory of Emotions 

Psychologists have wrestled with defining and conceptualizing 
emotions for some time (see, for example, James 1890; Reymert 1928). 
These issues are not yet resolved, but some major views and theories 
have been outlined. Westbrook (1983) introduced the emotion literature 
to CS/D research by adopting Izard's (1972; 1977) conceptualization of 
emotions. 1 This perspective has been used elsewhere in consumer 
research (Allen, Machle'it and Marine 1987) and also is the basic 
emotional theory for the dissertation (versus those used by others, 
e.g., Averill 1980; Mehrabian and Russell 1974). 



1 Other emotion models have been employed in consumer research 
(e.g., Havlena and Holbrook 1986; Peterson, Hoyer and Wilson 1986), but 
Westbrook provides the most lucid theory. 



82 



Izard defined a fundamental or discrete emotion (as opposed to a 
combination or pattern of emotions) as "a complex phenomenon having 
neurophysiological, motor-expressive and experiential components. The 
intraindividual process whereby these components interact to produce 
the emotion is an evolutionary-biogene tic phenomenon" (1977, p. 64). Of 
the dimensions mentioned in the definition, the motor-expressive and 
experiential components of emotion are the most relevant for 
understanding the emotional nature of satisfaction. 

Viewed within the perspective of Izard's framework, satisfaction 
would be considered a pattern of emotions. He defined a pattern to be 
"a combination of two or more fundamental emotions which under 
particular conditions tend to occur together (either simultaneously or 
in a repeating sequence) and to interact in such a way that all of the 
emotions in the pattern have some motivational impact on the organism 
and its behavior" (1977, p. 64). The fundamental emotions identified by 
Izard and a procedure for identifving which ones combine to constitute 
satisfaction are discussed below. 

One issue raised in the measurement section of Chapter 2 was that 
a major difference between attitude and emotion is the motivational 
dimension attributed to the latter. This aspect of the 
conceptualization of emotion is not unique to Izard, but rather has a 
long tradition in the literature dealing with emotions. Young (1961) 
suggested that emotions had come to be considered a subset of all 
possible motivations. The definition that he supplied reflects this, 
"an acutely disturbed affective state of the individual that is 



psychological in origin and revealed in behavior, conscious experience, 
and visceral functioning" (1961, p. 355). 

Taken together, the definitions by Young and Izard present a 
foundation for the conceptualization of satisfaction as an emotional 
state that is part of a more integrated process. Young states, "An 
emphasis upon the psychological situation in the origin of emotions is 
important because it is the 'psychological' origin that distinguishes 
emotions from other affective states" (1961, p. 346). That is, emotions 
are separate from feelings which originate from organic states (e.g., 
hunger, thirst) or relief from those states. 

Izard does not take so strong a position on this point. Although 
emotions and cognitions tend to interact, he proposes no strict 
unidirectional flow of causality (see also Zajonc 1980). Izard also 
postulates a discrimination between emotion and the cognition of 
emotion. That is, a person's ability to "objectively" label his or her 
feelings reflects a different phenomenon than simply reacting to those 
fee lings . 

On the other hand, Izard bases his measurement technique on an 
individual's ability to recognize not only the nature but also the 
magnitude of feelings. Izard's Differential Emotions Scale (DES) 
(1972; 1977) is, essentially, a list of adjectives representing the ten 
fundamental emotions. Subjects are asked to indicate on a five-point 
scale whether each word presented describes how they feel: 1) Very 
Slightly or not at all; 2) Slightly; 3) Moderately; 4) Considerably; or 



84 



5) Very Strongly. 2 The original (1972) version of the DES used 69 
adjectives to describe the ten emotions but shorter instruments have 
been used by Izard (1977) and Westbrook (1983). The ten emotions and 
the adjectives used in each of these later versions are presented in 
Table 4. The Item/Emotion correlations (factor loadings) that are 
reported were provided by Izard (1977, p. 126) to indicate reliability 
of the scale. 

Westbrook used his version of the DES to evaluate consumers' 
responses to their automobile. Using a factor analysis (1983) he 
identified four patterns of emotions. A subsequent cluster analysis of 
different data (Westbrook and Oliver 1984) yielded six patterns (Table 
5 lists the emotions that were most characteristic of each group, 
i.e., the dominant pattern). The latter analysis provides descriptions 
of the emotional profiles of specific segments of consumers and thus 
lends itself well to examining the definition of satisfaction as one or 
more of these profiles. 

The disconf irmation paradigm suggests that Surprise should be one 
emotion in the pattern defined to be satisfaction. Based en the 
Westbrook and Oliver (1984) analysis Cluster 2 and Cluster 4, which 
were characterized by the highest levels of surprise, would represent 
the dissatisfied and satisfied ends of a continuum, respectively (Eahr 
1982 also cites surprise as an important dimension of the 
dissatisf action--or disappointment--feeling. ) Contrary to the position 



2 For the purposes of the dissertation, only emotions as 
transient states are of interest. Thus, the DES II which measures 
emotional traits by collecting responses based on frequency rather than 
intensity is not discussed. 



Table 4 
Measures of Fundamental Emotions 



35 







Measures used 




Measures used 




by Izard Item/emotion 


bv Westbrook 


Fundamental Emotion 


1977 correlation 3 


1983 


Interest 


Attentive 


88 


Interested 




Concentrating 


79 


Excited 




Alert 


87 




Enjoyment 


Delighted 


81 


Delighted 


(Joy) 


Happy- 


87 


Joyful 




Joyful 


86 


Warmhearted 




Enthusiastic 






Surprise 


Surprised 


83 


Surprise 




Amazed 


85 


Amazed 




Astonished 


87 


Astonished 


Distress 


Downhearted 


86 


Sad 




Sad 


79 


Upset 




Discouraged 


82 


Discouraged 
Distressed 


Anger 


Enraged 


74 


Angry 




Angry 


84 


Bitter 




Mad 


86 


Provoked 

Rebellious 

Enraged 


Disgust 


Feeling of 








Distaste 


86 


Sickened 




Disgusted 


85 


Disgusted 




Feeling of 




Feeling of 




Revulsion 


78 


Revulsion 


Contempt 


Contemptuous 


89 


Sarcastic 




Scornful 


90 






Disdainful 


84 




Fear 


Scared 


88 


Fearful 




Fearful 


90 


Anxious 




Afraid 


89 


Inadequate 


Shame/Shyness 


Sheepish 


73 


Sheepish 




Bashful 


87 


Ashamed 




Shy 


88 




Guilt 


Repentant 


78 


Guilty 




Guilty 


83 


Blame v. or thy 




Blameworthy 


80 





Correlations are those reported by Izard (19" 



26' 



Table 5 
Patterns of Emotions in Automobile Ownership 



86 



a) Westbrook (1983; 



/estbrook and Oliver (1984; 



Factor I 

Anger 

Sadness 

Disgust 
Factor II 

Interest 

Enjoyment 
Factor III 

Shame /Shvness 

Guilt 
Factor IV 

Surprise 



Cluster 1 "Contented" 

Enjoyment 

Interest 
Cluster 2 "Mad as hell" 

Anger 

Distress 

Disgust 

Contempt 

Surprise 
Cluster 3 "Apprehension" 

Interest 

Fear 

Enjoyment 
Cluster 4 "Delighted" 

Interest 

Enjoyment 

Surprise 
Cluster 5 "Unemotional" 

all low responses 
Cluster 6 " "Guilty/Sheepish' 

Guilt 

Shame /Shyness 



taken by these authors, the remaining clusters are not considered to 
describe intermediate points of that continuum, however. Rather, these 
are considered to be emotion patterns which simply are not addressed by 
the satisfaction theory that is being explicated in the dissertation. 
This interpretation is consistent with the scores on satisfaction 
scales reported by the authors (for five scales with ten possible 
extreme—highest or lowest — scores, eight of the extremes were 
attributed to Clusters 2 and 4; all in the appropriate direction.) 
It should be noted at some point that the DES is necessarily a 
unidirectional scale. That is, the five-point scale which is used 
could be scaled only positively; negative "quantities" of emotions 
would not be interpretable by respondents or researchers. The ends of 
the satisfaction continuum are described by patterns of different 
emotions, but the emotions arise from the same process. Thus, although 
it may later prove useful to have discontinuous measures of 
satisfaction, such measurement should not be taken as an argument for 
separate constructs. 

Westbrook's treatment of the entire post purchase phase of 
consumer behavior under the satisfaction rubric ignores the necessity 
of engaging in a specific process to effect a state of satisfaction. 
That is, the present conceptualization explicitly outlines certain 
antecedent factors that are necessary to the arousal of satisfaction 
feelings. The process described by the antecedents clarifies the 
discrimination that theoreticians have sought to make between 
satisfaction and the lack of a satisfaction response (this issue is 
discussed in detail below). Westbrook's research procedure omits 



reference tc, and, in some cases, the presence of these factors. This 
omission makes further attempts to define the construct in terms of 
specific patterns of emotions problematic, especially since these were 
derived using his data. 

In addition, other aspects of the Westbrook studies raise concerns 
about basing the derivation of a measurement scheme on his findings. 
For example, the Differential Emotions Scale was modified but without a 
clear indication of the rationale of item selection or the necessity of 
diverging from Izard's own abbreviated scale. Westbrook' s implicit 
assumption that the modified scale will exhibit the same properties as 
the UE5 may not be borne out. Although these concerns do not condemn 
the data for the use to which they were applied, it is evident that the 
results are not adequate for the development of measures. Pilot Study 
I (Chapter 5) addresses these issues in more detail and is used to 
assess measures of the satisfaction construct which are based upon the 
DES. 

Motivation 

The importance of the distinction between satisfaction and 
non-satisfaction emotions lies in the notion that the motivational 
components of each of these classes of emotion are likelv to prompt 
different behaviors. The behaviors that are thought to result from 
satisfaction responses were identified in Chapter 2 and were even 
proposed as possible measures of the feeling. The validity of this 
measurement logic depends on how well the motivational relationship 
between the emotion pattern and the behaviors is defined. 



89 
Young describes motivation as "the 'energy' aspect of experience 
and reaction; a basic motivational principle is that varying degrees of 
stimulation liberate different quantities of energy" (1961, p. 16). 
Thus, the continuum that was presented in Figure 6 is actually 
reflecting this motivational concept of intensity (Atkinson and Birch 
1978; Mook 1987; Steers and Porter 1975). Moreover, it is more 
consistent with Young's treatment of motivation and emotion and with 
Izard's measures of strength of emotion to assume that the emotional 
continuum that underlies the behavioral responses reflects changes in 
response strength (Figure 7b) rather than changes in composition 
(Figure 7a). Weiner (1980) presents similar arguments in an 
interpersonal context. 

Finally, it is useful to note that motivation is a familiar 
concept in consumer research. CS./D research has borrowed a number of 
concepts from the dissonance literature, among these motivation. 
Dissonance was typically treated as a tension which people were 
motivated to reduce (Cummings and Venkatesan 1976; Oshikawa 1969). 
Although reduction was sometimes effected through choice of subsequent 
behavior (Cohen and Goldberg 1970), dissonance was typically assumed to 
initiate new behaviors (Kiesler and Pallak 1976). This discrimination 
is similar to the selection versus persistence effects of motivation in 
general (Atkinson 1957; Vroom 1964). As presented in Figure 5, exactly 
parallel decisions are thought to follow satisfaction or 
dissatisfaction when it occurs. In fact, it has been argued that 
arousal may be sufficiently non-specific that the individual is unable 
to label the response CZillman 1978). 



90 



Apprehensive 



Unemotional 
Angry Guiltv 



Surprised 
Contented 



Minimum 



Satisfaction 
(Likert Scale) -- 



Maximum 



a) from Westbrook and Oliver (1984) 



Low 



High' 



Interest 
Joy 



'Surprise 
Anger 
Distress 
Disgust 

.^Contempt 



Interest\^ 

Joy ^>High 

Surprise..^ 

Anger ^ 
Distress \ 
Disgust )>Low 
Contempt/^ 



Minimum Satisfaction 

b) based on Strength of Pattern of Emotions 



Maximum 



Figure 7 
Satisfaction As Emotional Continua 



Carlsmith and Freedman (1968) present an interesting discussion 
that linked dissonance and satisfaction (although they did not use the 
latter term). In essence, they indicate that dissonance arises when a 
disconf irmation can be attributed to internal factors ("I should have 
known better"). External factors ("You did not tell me that") are more 
likely to instill what has come to be called satisfaction. While this 
is compelling (see the attribution discussion in Chapter 2), most CS D 
researchers have assumed relatively complete information availability. 
Lack of a Satisfaction Feeling 

The discussion of motivation returns us to an issue that was 
briefly alluded to in the measurement section of Chapter 2; the 
occurrence versus non-occurrence of satisfaction. This question has 
been raised repeatedly since the earliest CS/D literature focusing on 
theory development (Day 1977; Hunt 1977a; Olander 1977). Empirical 
reports have been much less likely to acknowledge and control for such 
contingencies in descriptions of research designs and measurement 
procedures. This section of the dissertation integrates theoretical 
explanations of the occurrence of satisfaction with research designs 
and measures more sensitive to its non-occurrence. 

Given the current lack of detail within the disconf irmation 

paradigm, it is difficult to maintain a firm concept of how or when 

satisfaction occurs. It is not unusual to find theory and measurement 

techniques that are essentially in opposition on the issue. At one 

point, Day argued eloquently for added sensitivity to the detection of 

the feeling: 

It seems more plausible to assume that the consumer usualiy 
enters the consumption experience without even thinking about 



being satisfied or dissatisfied. In general, something out 
of the ordinary must occur either prior to the purchase, 
during the purchase process, or during consumption to alert 
the consumer or call attention to some aspect of the purchase 
situation. Research procedures which assume that an 
evaluation always takes place may produce one as an artifact 
of the research'." (1977, p. 150) 

Yet, only a short time later, he reported forcing satisfaction 

responses where, in fact, they may not have occurred. 

It was found that a five-point scale with a neutral midpoint 
appeared to provide a ' ccpout ' for a small number of 
respondents. When respondents were not offered a neutral 
response they seemed comfortable with making a choice between 
'somewhat satisfied' and 'somewhat dissatisfied'. (Day and 
Bodur 1979, p. 184) 

Swan and Trawick (1979) reported similar discomfort with their own 

methodology: "As an important methodological implication, we found that 

if you ask for an evaluation you will get one . . . Conceptually, if 

satisfaction is the product of a conscious evaluation, it follows that 

the research must first determine if an evaluation occurred before 

asking the consumer about his satisfaction" (p. 60). Comments such as 

these and acknowledgement of the issue in reviews of the CS/D 

literature (e.g., Dickson and Wilkie 1978) are laudable for raising an 

important issue but do not advance our understanding of the phenomenon. 

Some authors, on the other hand, have attempted to explain and 
predict the occurrence or non-occurrence of satisfaction. The 
Two-Factor theory discussed earlier posited separate causes of 
satisfaction and dissatisfaction. This implicitly theorizes a state in 
which neither occurs, although the techniques typically used (e.g., 



93 



critical incident) tend to avoid such situations (Maddox 1981; Swan and 
Combs 1976). 

A more recent cperationalization of the disconf irmat ion paradigm 
suggested that although performance is appraised against norms in all 
consumption situations, the discrepancy may not always be detectable. 
When performance falls in this "zone of indifference", no satisfaction 
feelings result (Woodruff, Cadotte and Jenkins 1983a). The width of 
the zone is postulated to vary among consumers and situations, although 
precise determinants of this dimension are not hypothesized. The 
authors propose that the factors which influence the width of the zone 
be studied and further suggest that unobtrusive measures may be 
necessary to avoid demand artifacts. 3 

Thus, the measurement techniques discussed in Chapter 2 gain added 
importance. If, as Day (1977) suggested, the typical "cognitive" 
measures of satisfaction produce measurement artifacts, then the less 
obtrusive emotional and behavioral measures should be considered. 
Behavioral measures were argued to be the least obtrusive (i.e., they 
can be collected by observation) but also the least sensitive (i.e., 
motivation must be relatively high to generate any response). The 
emotional measures proposed by Westbrook (1983) appear to be a 
reasonable trade-off between obtrusiveness (i.e., still a paper and 
pencil methodology, but with a less obvious research intent) and 
sensitivity (i.e., the responses are scalable). 



3 Note that these authors implicitly assume that only 
disconf irmat ion will produce a satisfaction response. As adapted in 
the dissertation, the disconf irmation paradigm does not make so strong 
an assumption. Given the appropriate circumstances, satisfaction can 
result from a confirmatory experience. 



Westbrook (1984) recognized the necessity of establishing the 
presence or absence of "true" satisfaction: "The study fails to provide 
a convincing answer as to whether indeed the presumed sentiments of 
satisfaction actually took place, and were not simply the product of 
the researcher's instruments" (p. 311). Yet, his use of the DES (1983; 
Westbrook and Oliver 1984) does not reflect a similar concern with the 
distinction. 

Based on the research reported by these authors, positive 
satisfaction can be defined tentatively as characterized by the 
presence of a pattern of emotions including Interest, Joy, and 
Surprise. Similarly, negative satisfaction can be characterized by a 
pattern including Anger, Distress, Disgust, Contempt, and Surprise 
(Figure 7b). This definition should not be interpreted as a revival of 
the separate constructs view of satisfaction; no difference in process 
is proposed. 4 

The satisfaction feeling as defined is considered to be a 
continuous range of emotional patterns in which the middle range is 
defined by a pattern of emotions that combine positive and negative 
elements. As one's feelings approach the ends of the continuum, both 
valence and strength increase which affects the increase in motivation 
that was cited earlier. The absence of any satisfaction feeling 
(positive or negative) is defined by patterns of emotions that lie 
outside the continuum. 



4 This definition, again, is tentative. It is based on studies 
which were not conducted with the objective of constructing a 
definition; Westbrook 1983; Westbrook and Oliver 1984. 



95 



Finally, it is appropriate at this point to recognize that CS/D is 

not the only literature to wrestle with this issue. Over the years, 

the questions of whether attitudes are as intense as had been presumed 

(Guttman and Suchman 1947) or even present at all (e.g., Converse 1970; 

Lastovicka and Bonfield 1982; Pierce and Rose 1974) have been raised 

and debated. Again, the arguments arise from the methodology that is 

typically employed in attitude studies. 

In that period I had been struck by two facts in particular. 
One was that very many respondents could not understand that 
a battery of pure opinion items had no objective 
'right-wrong' scoring, or that don't know responses were not 
a confession of the most abject ignorance, tc be avoided at 
all cost. The other was the frequency with which respondents 
chose a response alternative dutifully but accompanied the 
choice with side cues--shoulder-shrugging, eye-rolling, 
giggles and even sotto voce comments— indicating that they 
were very much out of their element and would pick any 
alternative haphazardly by way of helping me out. (Converse 
1974, p. 650) 

Lastovicka and Bonfield (1982) also noted this problem with demand 

artifacts and proposed that less leading measures (e.g., open-ended 

questions) be used to assess attitude or opinion (see also Heller 1982; 

Norpoth and Lodge 1985). Of course, difficulties arise in quantifying 

such techniques and estimating their reliability. But in some 

applications, these are less problematic than the over-estimation cf 

attitude presence that is likely to accompany forced-response 

questionnaires. Nevertheless, such questionnaires have dominated the 

attitude as well as CS/D literatures. 

Other Types of Satisfaction 

Th''s existential question is not limited to the consumer domain. 

Consumer satisfaction is not thought to be a unique feeling, but rather 

one manifestation of a more general pattern of emotions. However, the 



pattern of emotions is aroused only under certain circumstances, such 

as described by the enabling conditions that are described below. 

Various literatures have identified other targets of a satisfaction 

feeling which may be addressable under the paradigm presented here. 

Locke (1967) proposed a general satisfaction "affect", for example, 

that is cited in both the job satisfaction and CS/D literatures. 

These similarities have been noted by others working in the 

consumer domain. Westbrook and Oliver (1981) derived satisfaction 

measures from other literatures (e.g., marital satisfaction, life 

satisfaction) and theories of "consumer satisfaction" have been adapted 

directly from the job satisfaction literature (Adler and Robinson 1980: 

Maddox 1981; Swan and Combs 1976; Westbrook and Reilly 1983). Belk 

quotes William James thusly, 

A man's self is the sum total of all that he CAN call his, 
not only his body and his psychic powers, but his clothes and 
his house, his wife and his children, his ancestors and 
friends, his reputation and work, his lands, and yacht and 
bank-account. All these things give him the same emotions. 
If they wax and prosper, he feels triumphant; if they dwindle 
and die away, he feels cast down--not necessarily in the same 
degree for each thing, but in much the same way for all. 
(1986, p. 6) 

The job satisfaction literature dates back at least to the 1 930 ' s 

and, as such, is much more extensive than the research in CS/D (Locke 

1969). This has also allowed deeper development of individual theories 

and constructs than has thus far been the case in the consumer domain. 

One such theorv, Herzberg's Two-Factor theory (Herzberg, Mausner and 

Snyderman 1959), has been discussed as the antecedent to Two-Factor 

approaches in CS/D. 



97 



Although the specific contention that the attributes that cause 
satisfaction are separate from the ones that cause dissatisfaction has 
been rejected in Chapter 2 and has been discounted strongly even in the 
job satisfaction literature (Carroll and Tosi 1977), the general 
approach has been shown to be consistent with the disconf irmation 
paradigm as operat ionalized in the dissertation. That is, the emotions 
(similar to Locke's conceptualization) are evoked by disconf irmation of 
certain expectations about individual aspects of a job. The critical 
incident research technique suggests that Herzberg and his colleagues 
were, in fact, more concerned with specific disconf irmations than with 
general attitude toward the job. 

In contrast, the majority of the job satisfaction literature has 
focused on more deeplv ingrained, longer-held responses that seem more 
consistent with the attitude construct than with satisfaction (e.g., 
Steers and Porter 1975), as each was defined in Chapter 2. Typical 
measurement techniques consist of multi-item scales for rating various 
attributes of one's job (Smith, Kendall and Hulin 1969; Wanous and 
Lawler 1972). Recent organizational behavior texts also define job 
satisfaction in terms of attitude (Feldman and Arnold 1983; Reitz 1981) 
and report that the relationship between satisfaction and presumably 
resultant behaviors (Absenteeism and Turnover— equivalent to Loyalty 
and Exit in Figure 6) is weak. This apparent discrepancy between 
theory and findings may be the result of mis-specification of 
motivations. Vroom (1964), in a landmark book dealing with job 
satisfaction and motivation, made the distinction between motivation to 
act and motivation to act in a particular way. He chose to concentrate 



98 



on the latter and operationalized satisfaction with job as an attitude 
construct. The discrimination made seems to have been omitted in much 
of the later research that attempts to predict satisfaction responses 
from measures of the attitude. 5 

The Relationship Between Satisfaction and Attitude 

In Chapter 2, satisfaction and attitude were defined as separate 
constructs that should not be confounded by theory or measurement 
technique. However, the willingness of some authors (e.g., Evans 1983; 
LaTour and Peat 1979) to treat satisfaction as synonymous with 
postpurchase attitude reflects the strong links that do exist between 
the concepts. 

As an emotional response, satisfaction is experienced for a very 
limited time. Oliver (1980a; 1981) has recently proposed that the 
response does not merely dissipate, but rather "decays into" overall 
attitude. In the process, one's prior attitude is "updated" by this 
new experience. The process by which satisfaction decays and what 
factors cause or mediate the process are net clearly specified. Oliver 
(1981) hypothesizes an opponent process that exists to keep emoticnal 
levels within some "normal" homeostatic range (based on the work by 
Solomon and Corbit 1974). A theory such as this, however, does not 
suggest why attitude is affected but only that satisfaction will 
dissipate . 



- Satisfaction with marketing related posit ions--salesman, 
etc. --also has been discussed in the marketing literature as an 
attitude phenomenon, Becherer, Morgan and Richard 1982; Churchill and 
Pecotich 1982; Futrell and Parasuraman 1984. 



9° 



The short-lived nature of the satisfaction feeling is an important 
factor in the study of satisfaction effects. For example, if 
complaining is influenced by satisfaction but not by attitude as some 
findings suggest (Bearden and Teel 1983), then the motivation to engage 
in such behavior may dissipate along with the emotional pattern. 
Indications that this may occur arise from surveys concerning 
dissatisfaction and complaining behavior in which "I wanted to do 
something about it but never got around to it" is offered as a reason 
why a consumer did not complain (Day and Ash 1979). That is, there may 
be a discrete interval after a satisfying experience in which the range 
of satisfaction-related behaviors may result but after which the net 
effect is witnessed only through attitude. This is similar to the 
discrete interval in which mood seems to affect other social behaviors 
(Isen, Clark and Schwartz 1976; Walster .1964). This "window" may be 
regenerated in further experience with the item or in conjunction with 
related behaviors such as actual complaining (Andreasen 1977b; Diener 
198C). (On the other hand, as argued previously, certain behaviors 
such as repurchase are thought to be related more to attitude than to 
satisfaction — e.g. , Bearden and Teel 1983; Oliver 1977.) 

The relative influence of attitude and satisfaction on consumer 

behavior, as proposed, is depicted in Figure 8. The representation of 

the horizontal axis, Time, assumes that there is already some prior 

attitude which is affected only by the consumption experience at time 

t (the satisfaction occasion). The vertical axis, Resulting Behavior, 

is the continuum from Figure 6 rotated on one end. The figure shows 

that in the absence of a particular satisfaction occasion 
Resulting Behaviors 



100 



Compliment 



Positive 
Word-of-Mouth 



Loyalty 



Exit 



Negative 
Word-of-Mouth 



Complaining 



Emotion or 
Satisfaction 




Attitude 



>-Ti 



^/ 




Emotion or 
Satisfaction 



Stronger Influence on Behavior 
Weaker Influence on Behavior 



Figure 8 
Relative Influence of Satisfaction and Attitude 
on Certain Consumer Behaviors 



101 
(i.e, prior to t ) attitude guides behavior in the middle range of the 
scale. When satisfaction emotions are aroused (either positive or 
negative, as represented by the mirror image) the more extreme 
behaviors may be motivated. However, the motivation decays until 
attitude is again dominant ( t n ) and eventually there is no residual 
motivation. 6 

Specifying the nature of the reaction of satisfaction over time is 
an issue of critical importance to the researchers interested in the 
construct. Studies which are designed to examine satisfaction must 
measure the construct within the appropriate period or at least (as 
with Herzberg's critical incident technique) attempt to target the 
interval with the procedure and measurement techniques. Within the 
CS/D literature, studies typically have neglected this time dimension 
and created unnecessary ambiguity in their interpretation (e.g., 
LaBarbera and Mazursky 1983; Westbrook 1983). 

This confusion persists in part because satisfaction and attitude 
are related (as suggested in Figure 3), but the precise nature of this 
relationship has not been documented nor even well-conceptualized. 
Oliver's "decay" theory has sometimes garnered only weak support 
(Oliver and Linda 1981) and does not contain much explanatory power. 
Kennedy and Thirkell (1982) considered only the cognitive aspect of 
satisfaction and suggested that attribute level disconf irmation may 
increase the perceived importance or salience of that attribute which 



5 The decay of motivation does not stop at t n because there is no 
motivational component of attitude hypothesized. In fact satisfaction 
and attitude processes may be simultaneous but separate, as noted 
below. The figure however, is not limited to theories utilizing this 
argument, it serves merely to represent observed change over time. 



102 
will persist through future attitudes. Similarly, Fishbein and Ajzeri 
(1975) had proposed, in the more general case, that satisfaction and 
attitude processes occur in parallel. Rather than the decay of 
satisfaction causing attitude change, this theory posits the decay of 
satisfaction leaving attitude change as the sole trace of separate 
processes. Olson and Dover (1976) report changes in cognitive 
structure resulting from product trial that are consistent with this 
formulation (i.e., the beliefs about product attributes were different 
after usage experience). 

Discrimination and selection among these theories is well beyond 
the scope of the dissertation. However, the general framework provides 
for more discriminable conceptualization and measurement of the two 
constructs. The temporal relationship suggested is fairly well 
established (Westbrook and Cote 1980 found little evidence for a direct 
effect of prior attitude on satisfaction) and the measurement 
techniques that have been discussed are useful tools for the future 
examination of the process. The next section introduces a 
conceptualization of the general satisfaction process that posits 
certain antecedent conditions which must occur to evoke this more 
emotional, motivational feeling (as contrasted with simple attitude). 
The Extended Conceptualization 

Day's model of the satisfaction process (Figure 3) is an adequate 
representation of the CS/D literature at the time the model was 
published (1983). A revised version (Figure 9) incorporates much of 
the preceding theoretical discussion defining the construct and its 
properties . 



103 





to 
















a) 












(0 












c 












o 




CD 








a 




•a 






( 


ta 




3 








CD 




4-> 








w 




■H 
■P 
4J 








^» 














< 








c 










;i 










o 










•H 










-P 










u 












fl 












4-1 








■t-> 




CO 




CD 




u 




•H 




T3 




3 




4J 




3 




P 




CO 




4J 

•H 
4J 
4J 


-4-1 

c 

CO 


to 

c 


c 


/ 


■ 






o 

c 





















1 — 1 




u 


c 








■H 
10 




c 
\. o 

C -H 






U 


c 




! 










c 




— 


U 






o 




CO 


u 


CD 




O P 




•H 




_C 


l+H 


Q 




•rH (0 




*J 




*-> 


to 






p e 




CU 




>, 


•tJ 








fO l-l 


c 1 




p 


(ti 






S-H 

P 4-1 


u 






w 






1 


p 




T3 








•H C 




CD 




CU 


"it 






4-1 


1 


CU 












' — ; 




TJ 






o tn 


1 




cfl 


c 






U -H 


l 






CO 






Q 










CO 
t/2 

o 


o 




1 


CD 

u 

c 


0) 






* 








*--J 


CD 
•H 




CO 


£<~ 




1 


l-i 




u 


3 ° 






CD 






M C 












X 
















I 




bu c 






CO 




C 




4J 

0) c 




w 



















o 




CO (1) 






J3 


,*H 


•H 

■P 




c g 
O (1) 




CD 
T3 


■o 

CO 


5 


(0 




a> 




3 


—j 


P 




co <-i 




P 




a 


u 




CD o 




•H 


l-H 


CO 


CD 




« > 




■P 


>> 

> 


u 


a 




c 




P 


c 


X 

w 




M 




< 


(H 


c 
u 






' — m~ 


CO 












— 




cu 
en 

P. T3 


J= s 

C a2 


CO 
X! 
C 

CO 




O CU 


.-H N 


X 




■H r-t 




w 




Oi O 


to ^— ' 
u a 


c 




c 


3 +j 


< 




« 


+■> CO 
























m n 





U u 



!G4 



In contrast to the Day model, the new conceptualization describes 
a generic satisfaction process that operates to evoke the critical 
pattern of emotions in many situations. Thus, although Day strove to 
integrate all of the CS/D literature, the current model spans CS/D, job 
satisfaction (Locke 1969), marital or relationship satisfaction 
(Gurwitz and Dodge 1977; Rusbult 1983; Rusbult and Zembrodt 1983), 
interpersonal satisfaction (Brewer 1977), satisfaction with decisions 
(Kourilsky and Murray 1981), and satisfaction with particular types of 
consumer situations (leisure activities — Bloch and Bruce 1984; Sobel 
and McGuire 1977; patient satisfaction with healthcare--Swan and 
Carroll 1980; or satisfaction with legal services — LaTour 1978). 
The new conceptualization can be summarized as follows: 
Satisfaction is the feeling that results from (1) 
Expectations that lead to (2) a Decision that is 
followed by (3) personal Experience in which the 
characteristics of the object of the decision are 
subject to (4) Perception which, in instances of 
sufficient (5) Response Involvement , allows for a 
(6) Conf irmation/disconf irmation of the 
expectations and decision concerning the object. 
Note that the constructs are in no way labeled as specific to the 
consumer domain but can be operationalized in a variety of contexts. 

There are several features of the conceptualization worth noting 
here: (1) More constructs are proposed than in Day's (1977) model. 
Although the disconf irmation paradigm deals well with the cognitive 
dimension of satisfaction, more factors are necessary to understand the 



105 



total process. Each of the constructs is discussed in the following 
sections. (2) The six antecedent conditions are considered to be 
necessary and collectively sufficient for the arousal of the patterns 
of emotions characteristic of satisfaction. As such, the absence of 
any of these raises questions about the validity of any attempt to 
invoke the satisfaction construct. Thus, the conceptualization can be 
used as a template for the design of research to examine any type of 
satisfaction. (3) As with Day's model, the conceptualization is not 
meant to function as a causal model of the satisfaction process. The 
relationships designated in Figure 9 and the passage of time (denoted 
by movement to the right) are merely representative and do not 
encompass all of the influences that are possible. Derivation and 
validation of such causal theories will fellow demonstration of the 
necessity and collective sufficiency of the constructs. 

In the following sections, each antecedent condition is discussed 
in more detail. The discussions will generally be focused in the CS/D 
literature to allow clarity without loss of generality. Three of the 
constructs, expectations, perception and conf irmation/disconf irmation 
are well established in the dominant paradigm and are explicated in 
less detail than the three constructs that are being introduced. An 
evaluation of the role of this latter group (especially Decision and 
Response Involvement) constitutes the empirical portion of the 
dissertation. 



106 
Expectations 

A concept that has been central to the formation and use of the 
disconf irmation paradigm is the presence of prior expectations about 
the outcomes possible from a given situation. Comparison of the 
perception of actual outcomes against these expectations yields the 
disconf irmation judgment. Without expectations, satisfaction would be 
based solely on, and possibly would be indistinguishable from, 
perception. 

Within the CS/D literature, the source, nature and influence of 
expectations have enjoyed a prodigious amount of research attention. 
Cardozo's (1965) seminal paper assumed that expectations influence 
satisfaction and proposed an interaction between expectation and 
effort. Later findings suggested that expectations may even affect 
perception by masking reality or, at least, subjects' reports of their 
perceptions of reality (Anderson 1973; Anderson and Hair 1972; Deighton 
1984; Olshavsky and Miller 1972; Olson and Dover 1978). Some of the 
results reported in Chapter 6, below, support this contention. 

This attention to the expectations construct has not waned in 
recent years. Broad models of the entire satisfaction process continue 
to center on the disconf irmation of expectations; and calls for better 
understanding of the construct are common (Day 1983; Woodruff, Cadotte 
and Jenkins 1983a). The problem with much of the research that 
attempts to address these issues is that it is not conducted within the 
context of these more complete theoretical networks, leading to results 
that may be bound by situation. Nevertheless, the concept of 
expectations has become axiomatic in theories of individual 



107 



satisfaction and is used to account for levels of satisfaction in 
macro, societal contexts (e.g., Summers and Granbois 1977). 

Less well agreed-upon is a specific definition of expectations. 
Among those offered have been 

1) "the perceived likelihood that a product possesses a 
certain characteristic or attribute, or will lead to a 
particular event or outcome" (Olson and Dover 1976, p. 169) 

2) "pre-decisional or pre-consumption beliefs about the 
overall performance of the product created by manufacturer's 
claims or test-report information" (LaTour and Peat 1979, 

p. 433) 

3) prior attitude (in Fishbein's affect-belief sense) 
representing a homeostasis position prior to adaptation to a 
new experience (Oliver 1981) 

4) expectations of satisfaction (Bahr 1982; Oliver 1981) 

5) predicted performance of a brand (versus a performance 
norm within product class) (Woodruff, Cadotte and Jenkins 
1983a) 

6) "An expectation of product performance is simply a 
prediction of what performance will be . . . expectations 
[and] norms . . . are also measured for the social benefits 
of the consumption event and the total economic costs of the 
consumption event" (Day 1983, p. 113). 

A number of issues are unresolved in the treatment of expectations 
within the CS/D literature, as evidenced by the above definitions. 
Whereas earlier conceptualizations focused on attribute level 



cognitions (e.g., Olson and Dover 1976), more recent theories have 
tended to operate at the overall perception level (Day 1983) and seem 
to approach the type of utility expectation discussed by Peter and 
Tarpey (1975). 

Furthermore, the limitation of expectations to simple predictions 
about the future is also not unanimously accepted. A recent review 
(Gilly, Cron and Barry 1983) of the research that focused on the 
expectations portion of the model pointed out that investigation in 
this area is still active. Of the four types of expectations suggested 
by Miller (1977), Expected, Deserved, Ideal and Minimum Tolerable, only 
two have received much support as discrete constructs. "Expected" 
expectation is essentially the probabilistic prediction most often 
referred to in satisfaction theories in both the consumer and job 
literatures (e.g., Filer 1952; LaTour and Peat 1978; Summers and 
Granbois 1977; Swan, Trawick and Carroll 1981; Wanous and Lawler 1972). 
Deserved expectations, on the other hand, are more of an equity 
phenomenon, based on the individual's "input" to the system (e.g., 
Sobel and McGuire 1977; Swan and Trawick 1980a). Although measures 
taken can discriminate between these types, no real difference in 
process is assumed. 

The other two types of expectations proposed by Miller, Ideal (the 
"wished for" level — similar to the aspirational level posited by Filer 
1952) and Minimum Tolerable (absolute "bottom performance"), have not 
been discriminably measured, nor is it clear in the present 
conceptualization how they might affect satisfaction separately (Gilly, 
Crcn and Barry 1983; Sobel and McGuire 1977). It seems more likely 



109 



that all of these types of expectations combine to form an "Expected" 
expectation or simple prediction, as described by Day, that is 
reminiscent of a Bayesian probability judgment (Winkler 1972). 

Day (1983) and Woodruff, Cadotte and Jenkins (1983a, b; Cadotte, 
Woodruff and Jenkins 1983) treat expectations as a brand (or possibly 
item) specific construct. That is, the predictions made concern the 
specific target of the satisfaction feeling. At the more general (say 
product class) level, prior experience will have generated judgments 
about the normative perception for that situation (Cadotte, Woodruff 
and Jenkins 1987; Chase and Ericsson 1982; Chase and Simon 1973). 
These norms are presumed to have an influence separate from 
expectations on disconf irmation (Figure 3). This possibility is 
related to the complicating effects of expectations on relevant 
schemata (Alba and Hutchinson 1987) but the effect on CS/D constructs 
has not been demonstrated empirically, as yet. 

Finally, other "types" of expectations have been identified and 
discussed. One segment of the CS/D literature deals with Comparison 
Level theories of satisfaction. In essence, these propose that 
disconf irmation results from evaluating obtained perception in light of 
a Comparison level that is composed of predictions based on (1) one's 
prior personal experience (2) experiences of relevant referent persons 
and (3) unique characteristics of the present situation (LaTour and 
Peat 1978, 1979; Miller 1977). Although this is sometimes presented as 
a competitive explanation to disconf irmation (Barbeau and Quails 1984; 
Haines 1979; Swan and Martin 1981), the structure is consistent with 
the formation of a Bayesian probability judgment. In effect, the 



1 10 
so-called Comparison Level formulation is more a theory of the source 
of expectation than of its structure. Thus, this is better discussed 
with other theories of expectation formation, such as those dealing 
with advertising and interpersonal communication (Gilly, Cron and Barry 
1983; LaTour and Peat 1980; Oliver 1979) than as an explication of 
separate constructs. As suggested by Gilly, Cron and Barry (1983),- 
this attention to the source of expectation may reveal the process by 
which expectations change over time, thus creating different levels of 
satisfaction over time. 

The concept of expectation that is accepted for the dissertation, 
then, is essentially the adaptation level proposed by Oliver (1981), 
operationalized in a manner similar to Day (1983). In contrast to 
Day's treatment, however, the more general norms that he proposed be 
kept separate are believed to be encompassed within the prediction. 
This probability judgment results from a Bayesian-type process and can 
be generated for a target stimulus at any level of specificity (Table 
2) that is within the person's experience. (Day and Woodruff et al. 
focus on brand or item satisf action--making it difficult to extrapolate 
beyond the concept of brand expectations and product class norms.) 
Within the present model, separate attribute level expectations are not 
thought to be relevant to the net effect of the satisfaction process. 
As will be discussed in the "Experience" section, salience of specific 
attributes may change through consumption or a discrepancy on an 
individual dimension may arouse a higher "Response Involvement" level. 
At this point, we are concerned only with an overall expectation that 



11 1 



results in a decision to commit to the stimulus that may become the 

target of a satisfaction feeling. 

Decision 

In the CS/D literature, satisfaction historically has been 
discussed as a "postpurchase" phenomenon. But rather than assume that 
this label merely designates a passage of time, it is argued that the 
purchase act itself is necessary for and can exert an influence upon 
the satisfaction process. At the more abstract level at which the 
conceptualization is presented, the required act is described as the 
decision to make a commitment to the satisfaction target. 

A great majority of the research in consumer behavior examines the 
factors which result in a consumer's choice or selection of a good. 
There has been comparatively little attention paid to the outcomes of 
such choices. The literature that arose relating cognitive dissonance 
(Carlsmith and Freedman 1963; Festinger 1957, 1964) to consumer 
behavior, however, did indeed focus on the choice and its effects on 
the individual. In their review of this literature, Cummings and 
Venkatesan (1976) argued that certain prerequisite conditions must be 
fulfilled in order that dissonance occur. Among these were cited 
"volition and irrevocable commitment to the decision (product choice)" 
(p. 304). These conditions circumscribe that which is being called 
decision in the conceptualization under discussion. 7 

Some of the earliest papers that are cited often in the CS/D 
literature (e.g. Cardozo 1965; Cohen and Goldberg 1970) arose from this 



7 Cummings and Venkatesan also argued that the decision must be 
important. This is treated separately under Response Involvement, 
below. 



12 



dissonance tradition. Cohen and Goldberg specifically manipulated 

decision (among other manipulations), as defined herein, to create "no 

dissonance" control groups. Unfortunately, no checks of this 

manipulation were reported, so the success of the strategy can not be 

easily assessed. Nevertheless, decision was believed to act to 

solidify cognitions and make them more resistant to change, thus 

arousing dissonance when the prior cognitions were disconfirmed (see 

Cialdini, Cacioppo, Bassett and Miller 1978 for a similar discussion). 

The dissonance literature is not without its limitations and 

criticisms. Even fairly early in the stream, Davidson and Kiesler 

(1964) raised concerns regarding observation of dissonance or its 

effects prior to or without a decision being made (also, Carlsmith and 

Freedman 1968). The authors cited Festinger's sharp distinction 

between "conflict" (a pre-decision construct) and "dissonance" (a 

post-decision construct) . 

Janis (1959), on the other hand, thinking in terms of 
'conflict resolution,' implies that there is little or no 
distinction between pre- and post-decision behavior and that 
systematic re-evaluation occurs both before and after the 
decision. (Davidson and Kiesler 1964, p. 10) 

A study conducted with thirteen to sixteen year-old girls supported 

Festinger's dichotomy, albeit not conclusively. Directional evidence 

is found for a decision effect on re-evaluation of job candidates. As 

Festinger points out, however, a counter-explanation merely requires an 

inference that delay between instruction and decision was sufficient to 

allow re-evaluation. 

It is also not clear whether the early dissonance experiments 

forced a decision under circumstances in which the subjects would not 



1 13 



ordinarily decide or try actively to avoid a decision (Braden and 

Walster 1964; Weick 1967). This forced decision may counter, or at 

least alter, the effect of the decision on the resulting cognitive 

structure. 

Freedom to choose is the immediate precursor of 
responsibility, for interpersonal life holds one accountable 
for what he chooses, not, ordinarily, for what he is forced 
to do. (Temerlin 1963, p. 39) 

This free, volitional decision is conceptually similar to the 

commitment construct elaborated by Kiesler and labeled "behavioral 

commitment" by Johnson (1973). 

Kiesler defined commitment as "the pledging or binding of the 

individual to behavioral acts" (1971, p. 30). This definition does not 

reflect all of the complexity of the construct, however. Kiesler gives 

more clues to the nature of commitment by proposing experimental 

manipulations that presumably would vary the level of commitment. 

These include varying 

1) the explicitness of the act (of commitment) 

2) the importance of the act for the subject 

3) the degree of irrevocability of the act 

4) the number of acts performed and 

5) the degree of volition perceived by the person in 
performing the act. 

Similar manipulations of the decision construct proposed in the 
dissertation may also yield a complex structure. At this time, 
however, only a minimal structure is hypothesized. It is essential 
that an explicit, irrevocable, volitional choice be made (e.g., of a 
product, of behavior with respect to a product, etc.) in order that the 



1 u 



satisfaction feeling, as conceptualized herein, result. It is argued 

that one act is sufficient to engage the satisfaction process; 

importance of the act is discussed below under Response Involvement. 8 

Kiesler (1971; Kiesler and Pallak 1976) also argued that the 

result of "fixing" cognitions is central to his conceptualization of 

commitment (see also Aronson 1968; Festinger 1957; Gerard 1965, 1968; 

Lewin 1951, 1952). A similar case has been made in the marketing 

literature, 

Involvement results from the fact that important values or 
the person's self-image are engaged or made salient by a 
decision situation. This would lead to the arousal or drive 
activation that Mitchell (1979) has suggested. Commitment 
([which is approximately equal to] position involvement) 
results when these values, self-images, important attitudes 
and so on become cognitively linked to a particular stand or 
choice alternative. (Crosby and Taylor 1983, p. 415) 

Tentative support for this proposition was derived from an analysis of 

data regarding attitudes before and after a vote on a referendum. A 

high commitment group of voters exhibited greater stability in voting 

preference over time than did a low commitment group. In addition, 

post-election perceptions of outcomes were more highly correlated with 

pre-election preferences for the high commitment than the low 

commitment group. 

In Kiesler' s view, the strength of resistance to change (or amount 

of dissonance generated) is likely to be related to the level of 

commitment which can be varied by manipulating the factors cited above 

or other aspects of the situation. Effort is one such variable that, 



8 Another use of the term, commitment, with regard to consumer 
behavior mere closely approximates what is now called involvement which 
is discussed below — Robertson 1976. 



1 15 
as Kieslar stated, "should have a dramatic impact on behavioral 
commitment" (1971, p. 172) but its influence results from a process too 
complex to understand so early in the development of commitment theory. 
A similar argument might be raised in explaining a complex effect of 
effort on satisfaction (Cardozo 1965). 

For the purposes of the dissertation, various degrees of 
decisiveness and their impact on satisfaction are not discussed. At 
this point in the development of the conceptualization, it is simply 
argued that making a decision is a necessary antecedent to the 
initiation of the satisfaction process (see a related discussion by 
Weick 1967). In this regard, the definition of decision that has been 
adopted is consistent with the conceptualizations of early dissonance 
theorists who discussed separately the decision that was made and the 
importance of the situation in arousing dissonance (Brehm 1956, 1959; 
Brehm and Cohen 1962; Festinger 1957). 

It has been argued previously that the satisfaction process 
results in an emotional, motivational state which is similar to the 
motivation aroused by "dissonance inducing" processes (Kiesler and 
Pallak 1976). In a sense, the motivation in each case arises from the 
"ego involvement" (a concept separate from but related to Response 
Involvement, discussed later) that results from the necessity to make, 
and adhere to a decision (Filer 1952; Locke 1967). Others have 
suggested that this self evaluation phenomenon occurs and is motivating 
for the individual (see, for example, Calder 1973, p. 257). 

In the CS/D literature, one explanation that has been proposed for 
the function of expectations in determining satisfaction (versus a 



1 16 
Sxmple perceptual process) has centered on this justification of the 
consumption decision (Czepiel and Rosenberg 1977; Shapiro 1968). In 
essence, it is the decision that is disconf irmed, rather than the 
expectations (Carlsmith and Freedman 1968). This purchase concept is 
reflected in some models of the satisfaction process, but is more often 
ignored (Day 1983) or positioned off to the side, not directly 
impacting on the satisfaction result (Andreasen 1977a). 

The conceptualization of satisfaction being presented (Figure 9) 
models decision explicitly as a direct input to the disconf irmation 
judgment; expectations lead one to make such a commitment. Thus, one's 
ability to make the choice is relevant in determining the level of 
satisfaction which is ultimately experienced, a prediction that has 
some support from research findings dealing with the behavior of both 
individual (Folkes 1984a; Wes-tbrook 1980c) and collective (Lambert, 
Dornoff and Kernan 1977) consumers. Similar arguments can be raised 
from reports that individuals' self-concepts may be actively involved 
in consumer situations (Grubb and Grathwohl 1967; Sirgy 1982). That 
is, by committing to a product (or other "target"), the individual is 
risking his/her identity as a capable decision maker (Walton and 
Berkowitz 1985). Smith and Swinyard (1982), on the other hand, propose 
that trial rather than true commitment is the object of many purchase 
decisions. Their formulation seems flawed in that the decision appears 
to be between "buy versus don't buy" rather than one of "which 
alternative do I prefer?" It is this latter decision that is claimed 
herein to be the more common and, therefore, more central to the 
satisfaction process. 



i 17 

This proposal of decision as a crucial link in the satisfaction 

process is not new. Dickson and Wilkie (1978) raised specifically the 

issue that, without a decision, a postpurchase condition can not be 

present. Folkes (1984a, b), Russo (1979) and others have discussed the 

possibility of the decision to complain resulting from a process by 

which blame for a problem is attributed either to another (e.g., "He 

made me look bad, so I'll gripe to or about him") or to one's self 

(e.g., "I made a bad decision; I'll keep my mouth shut"). 

Admittedly, it is at times difficult to isolate the point at which 

a decision is made. Decision making by groups (e.g., families, 

industrial buying groups) may have emotional results for the individual 

members who feel the requisite volition was present. Less clear, 

however, is the case where the purchaser is different from the user. 

Many consumer behavior approaches to satisfaction presume 
that the consumer has voluntarily chosen the product in 
question. For example, the presence of free choice is a 
prerequisite to the arousal of cognitive dissonance. An 
interesting set of issues therefore arises when the consumer 
evaluates a product or service chosen by someone else. 
(Solomon 1986, p. 216) 

In circumstances such as this, the user would experience satisfaction 

if (1) there is sufficient volition in the choice to use the product, 

or (2) there was a decision made to accept the buyer's judgment 

(assuming the other conditions are also met). In the latter instance, 

the buyer acts as the user's agent and has had the (purchase) decision 

making authority delegated (Jensen and Heckling 1976; Ross 1973: 

Solomon 1986). The user still has a judgment at risk, sufficient to 

arouse emotion, but not the selection of the product. This can result 

in redefining the target of the feeling (e.g., from the product to the 



1 18 

buyer) and subsequent behaviors, but this issue is beyond the scope oi 
the dissertation. 

As defined so far, the pattern of emotions that comprises the 
satisfaction feeling does not include shame, guilt or other indications 
of personal responsibility (ego involvement). Although this would 
clarify the relationship with decision somewhat, it may not be 
necessary to incorporate these emotions. The scales used have, again, 
not been seeking a clear definition of satisfaction so that the 
omission of these internally directed emotions may be an artifact of 
the measurement procedure. It should also be noted that satisfaction, 
when acting as a precursor to ext ernally directed behaviors , may evoke 
less strong internal responses. Thus, how decision affects and is 
reflected in satisfaction and how the selection of subsequent behaviors 
is influenced are empirical questions that can not be answered 
convincingly at this time. 

The foregoing discussion, however, does make clear that decision 
is necessary to a "purchase" situation. Satisfaction that is measured 
after a typical marketing experience pre-supposes this purchase event 
(cf. Crosby and Taylor 1982). When satisfaction is studied in the 
behavioral laboratory, however, this step can be and has been omitted 
at times. For example, in Cardozo's (1965) study, subjects were never 
given an option as to the task they were to perform, the "reward" they 
were to receive, or any other aspect of the experiment. Thus, the 
effort manipulation was not likely to have the expected impact 
(involvement may have been raised, but emotions were probably not 
aroused) . 



1 19 
Similar to the Cardozo study, other early experiments dealt with 
factors influencing postpurchase evaluation and did not arouse 
satisfaction because there was no opportunity to make an explicit, 
volitional and irrevocable decision (Anderson 1973; Oliver 1977; 
Olshavsky and Miller 1972; despite their frequent citation as 
satisfaction studies. In some cases (e.g., LaTour and Peat 1980) it is 
suggested, but not clearly stated, that subjects may have chosen to 
participate in the study on the basis of a promised reward. The 
disconf irmation of such a choice may be adequate to drive the 
satisfaction process. 

More often, however, the necessity of demonstrating that purchase 
occurred prior to studying a "postpurchase" phenomenon is ignored. 
Even recent satisfaction research has neglected this issue (e.g., 
Churchill and Surprenant 1982; Wilton and Tse 1983). Some attempts to 
explain the effects of advertising messages that are counter to reality 
have invoked the disconf irmation paradigm and the satisfaction 
construct (Deighton 1984; Olson and Dover 1978). These, too, are 
predicated on the assumption that expectations are disconfirmed and 
ignore the opportunity to show that decisions were made based on the 
expectations . 

In summary, this section has argued that decision is a necessary 
antecedent to the satisfaction process. Aspects of constructs labelled 
"decision" and "commitment" in consumer behavior and psychology 
literatures have been merged into the single construct presented here. 
It does not seem fruitful to attempt at this time to disentangle 
similar presentations of differently named constructs (Fendrich 1967; 



120 

Hartley 1967). Expectations doubtlessly influence the direction taken 
by this decision, but are insufficient to arouse disconf irmation. What 
enters into the disconf irmation, then, is the individual's self-concept 
of his/her decision making ability plus a summary expectation of the 
result of that decision. Moreover, the expectation is now more 
resistant to further changes because, in effect, more is at stake. 
This expectation is compared, within the disconf irmation process, 
against a perception of the "true" result of the decision—which only 
can be derived from personal experience. 
Experience 

The point was made earlier that in instances in which the 
purchaser and user of an item are different people, it is the user (if 
either) who will experience a satisfaction feeling. This follows from 
the assumption that, whereas decision or commitment making authority 
can be assigned, the experience of personal "consumption" can not. 9 

In the general conceptualization, experience is defined to be 
sufficient interaction with the target (of the feeling) to allow 
evaluation of its characteristics (Fazio et al. 1982; Fazio et al. 
1986). In the usual paradigm within the CS/D literature this simply 
amounts to consuming enough of a product to characterize its 



9 The special case of satisfaction with gift-giving or donation 
is not addressed specifically here. Such situations can be treated 
under the same conceptual framework, but the problem of including 
expectations and perceptions with respect to the recipient's reaction 
is an additional complication which can be addressed better as a 
separate issue. 



12i 
performance. 10 Rather than a hypothetical construct, experience 
constitutes an (often) observable activity or event that leads to the 
satisfaction process being initiated (given the other necessary 
conditions). In contrast to cognitive dissonance (which focuses on the 
decision event), satisfaction requires information additional to that 
available in making the decision. Without this added information, 
there could be no disconf irmation. (Satisfaction is sometimes referred 
to as a postconsumption, rather than merely postpurchase, concept-- 
LaTour and Peat 1979. ) 

Surprenant and Churchill (1984) have raised the question of 
whether vicarious evaluation of product performance is adequate. In a 
comparison of groups that actually used a product versus others that 
were merely told of the product's performance, the latter had generally 
more compressed results (i.e., closer to mean responses). This finding 
corroborates prior arguments that subjects in role-playing experiments 
are likely to be less involved than those participating in more 
realistic situations (Sawyer 1977). The necessity of determining that 
personal experience has occurred prior to assessment of satisfaction 
had been noted in passing by some authors (e.g. Park and Bahr 1980); 
but in general, the issue is not raised. 

The importance of the problem can not be dismissed so lightly, 
however. The process by which an individual perceives an object, event 
or another individual and integrates that perception into knowledge is 



13 Within the dissertation, "experience" is used to refer to this 
consumption. In other contexts, "prior experience" has been used to 
refer to one's history with the item, product class, etc. prior to the 
purchase decision (Alba and Hutchinson 1987). 



I 22 



not well understood. Fazio et al. (1982) examine the influence of 
experience on attitude accessibility, a conceptually similar but 
discrete concept as outlined above. The early studies of postpurchase 
phenomena were attempts to identify the influence of expectations on 
the perceptual process (Anderson 1973; Olson and Dover 1975). In 
essence, it was believed that the information that could be derived 
from perception is somehow altered by expectations (LaTour and Peat 
1979). These theories depended on the assumption that the individual 
was free to draw conclusions from a given consumption situation. 

Occasionally, however, laboratory experiments or other studies 
that did not allow the opportunity for personal experience have been 
reported in the CS/D literature. Although it is not certain what the 
actual effects of such omissions are, there are suggestions from 
psychology as to how processing may be affected. For example, in a 
similar discussion of experience-based versus indirect (i.e., based on 
second-hand information) attitudes, Fazio and Zanna (1981) proposed 
that actual personal experience may not only make larger quantities of 
information available but may also change the focus of attention and 
attribution. Just as the previous section discussed the personal 
commitment required for satisfaction (i.e., risking one's decision 
making skill); experience challenges the individual's skill as a user 
("Did I cause the good/bad performance?). Folkes reported that 
consumers do at times make such attributions concerning the cause of 
product failures (1984b; also, Krishnan and Valle 1979; Valie and 
Wallendorf 1977) . 



123 
In addition to amounts of information available and types of 
attributions made, personal experience can influence how the 
information is processed further or what response is generated. Fazio 
and Zanna (1981) cite a study reporting similar attitudes but different 
behaviors found in samples of students who were either affected 
directly or only tangentially by a housing shortage. The literatures 
on decision making and judgment provide insights into which aspects of 
information affect its processing (Ofir 1982). Lyon and Slovic (1976), 
for example, reported that subjects prefer to base judgments on 
information that they perceive to be most specific to the judgment they 
are making and to ignore other available data that could result in a 
probabilistically better judgment. Essentially, they argued that 
specific information was more salient than more general, more abstract 
data. This process is demonstrated quite nicely in a consumer context 
in which a subject is seen wavering between objective information 
placed before her and information from prior personal experience 
(Smead, Wilcox and Wilkes 1980). 

Similarly, Kahneman and Tversky (1973) had reported that the 
relative representativeness of information affected its usefulness for 
individuals. In the current context, it might be argued that a 
consumer would find self-gathered information more representative of 
the specific item than information provided through other means. 
Bar-Hillel's (1980) discussion of relevance and specificity of 
information would seem to rule out even personal experience with a 
sample or exemplar that is not the specific target of the commitment 
(also, Kamins and Assael 1987). 



24 



In general, the process of gathering and evaluating information 
about a target object and generating a response is rather complex (cf. 
Zajonc 1968). The factors that influence what information is used and 
how it is used would seem to be sensitive to changes as gross as the 
omission of the actual object. Our knowledge is not yet so 
sophisticated that we can correct for this in experimental designs or 
analyses (Smith and Swinyard 1983). These arguments were suggested by 
Olson and Dover in an early report in which they concluded that 
personal experience, "may have a greater effect on the acquisition 
and/or change of cognitive structure elements than does information 
from sources such as advertising or word-of-mouth" (1976, p. 173). At 
the time, however those authors were making a distinction between the 
effects of expectations (based on the latter information sources) and 
disconf irmation on cognitive structure. They did not anticipate the 
potential use of indirect experience to generate disconf irmation. In 
fact, they reported that even, "usage experience with most consumer 
products may not create the extreme disconf irmation necessary to 
produce a contrast effect" (p. 173). 

The problem is not limited to the laboratory experiment. Locker 
and Dunt (1978), in their review of patient satisfaction in the United 
Kingdom, reported difficulty in determining whether surveys assessed 
satisfaction of consumers or only potential consumers. The surveying 
problem is further compounded by the as yet unknown effects of repeated 
purchases and/or experiences (Kennedy and Thirkell 1982; Lutz 1980) on 
measured satisfaction. With field surveys, such problems arise only 
when non-consumers attempt to respond or frequent consumers differ in 



125 
their timing of expectations and disconf irmation. If the sample is 
representative, then the extent of these problems should reflect the 
characteristics of the population. 

Justification of lack of personal experience in behavioral 
experiments is not so easy to provide. Sawyer (1977) argued 
convincingly that role-playing situations are not appropriate for the 
majority of experiments which purport to draw conclusions from the data 
(also, Orne 1962). Yet, consumers are frequently put in "What if?" 
situations and asked to provide rating scale responses (e.g., Churchill 
and Surprenant 1982; Jordan and Leigh 1984; Lambert, Dornoff and Kernan 
1977; Mowen and Grove 1983; Sirgy 1984; Winter 1974). In other cases, 
consumers are asked to respond to the experiences reported by strangers 
(Deighton 1984) or to report their own satisfaction prior to use of the 
object (Sirgy 1984). Day (19-83) excluded specifically this latter 
"'anticipated satisfaction" from his general conceptualization. 

As with the other concepts in the current model, experience can be 
defined at various levels of specificity. If satisfaction with a 
retailer is of interest, for example, then experience with that 
retailer--rather than with a specific product--is the requirement. 
Most situations for which consumers can readily volunteer satisfaction 
reactions or report that they have engaged in satisfaction-driven 
behaviors will have a readily apparent experience episode enveloped in 
the description. 

This section has argued primarily that experimental designs must 
include not only an opportunity for decision, but also the opportunity 
for personal experience in order that a realistic satisfaction process 



26 



may obtain. The most basic cause of this requirement is the 
investigator's inability to pre-specify all of the applicable 
information in a form usable for the subject to derive his perception 
of "reality" . 
Perception 

The CS/D literature has recognized since its inception that 
objectively defined products or product performance do not evoke the 
same response from all observers. A variety of factors can influence 
an individual's impression of reality. In essence, the following 
working definition holds, "Perception can be thought of as the 
individual's mental impression of a stimulus object" (Anderson and 
Hair, referring to product perception, 1972, p. 70). The studies that 
initiated interest in the postpurchase phase of consumer behavior 
(Anderson 1973; Anderson and Hair 1972; Cardozo 1965; Oliver 1977; 
Olshavsky and Miller 1972; Winter 1974) and are often cited as 
satisfaction studies (LaTour and Peat 1979) actually treated perception 
as the dependent variable. In effect, each investigation was more or 
less interested in the role that prior expectations played in altering 
one's impression of the performance of a product. 

A review of the literature dealing with the perception construct 
suggests that each of the preceding antecedent conditions, i.e., 
expectations, commitment and experience, may affect the individual's 
perception of the target (Kassarjian and Robertson 1973). Early CS/D 
research typically manipulated onlv two of these, expectations and/or 
experience (i.e., in this case, experience is analogous to the 
manipulations used to demonstrate good or bad product performance). 



27 



The conclusion that can be derived from these studies is that 
perception is largely the result of the performance that is obtained in 
actual experience, but can be significantly modified by expectations 
(especially if the expectations have been solidified by commitment). 

Recent theoretical treatments of satisfaction and the 
disconf irmation paradigm have focused on expectations and perception 
(Day 1983; Woodruff, Cadotte and Jenkins 1983b) although performance is 
still the most common operationalizat ion of the latter construct (e.g., 
Surprenant and Churchill 1984). In empirical research, perception 
dominates expectations as a cause of satisfaction (Churchill and 
Surprenant 1982; Sobel and McGuire 1977; Swan and Combs 1976) although 
each is usually confounded with disconf irmation (generally 
operationalized as the difference between expectations and performance/ 
perception). Perception is also thought to play a role in determining 
responsibility for a given situation occurring (i.e., self versus other 
perceived as a cause) that may influence complaining behavior 
(Kassarjian and Robertson 1973; Settle 1973). Further, perception of 
performance may impact directly on the definition of the individual's 
"evoked set" of alternatives (Narayana and Markin 1975), thus 
influencing repurchase without a satisfaction or attitudinal mechanism, 
per se. 

There are various explanations for why perceptions differ from 
objective reality. Choice of products and judgments of the relative 
preferability of attributes have been shown to be sensitive to the 
context in which the judgment is made, including the range of 
alternatives available (Assar and Chakravarti 1984; Chakravarri and 



128 
Lynch 1983; Olshavsky and Miller 1972) and there is some evidence that 
the presence of negative attributes or of negatively evaluated levels 
of attributes alters the manner in which items are evaluated (Lynch 
1984b; Sobel and McGuire 1977). As Hoch and Ha (1986) demonstrate, 
objective reality itself is dependent on the ambiguity of product cues. 
Some products, e.g., paper towels which differ on manufacturing 
specifications, are easier to compare than others, e.g., shirts 
differing only on color (See also Bergh and Reid 1980.). 

Anderson (1973) described in some detail the more comprehensive 
theories of judgment that may have implications for perception. (1) 
Discrepant performance experiences are assimilated closer to positive 
expectations. In effect, being wrong is uncomfortable (here, it seems 
to be the decision being disconf irmed) and the individual reduces this 
discomfort by altering his sense of "reality" — essentially a dissonance 
explanation (Festinger 1957). (2) A somewhat similar conceptualization 
was presented as a generalized negativity theory. In this case, any 
inconsistency with expectations (good or bad) causes the discomfort 
that people seek to reduce. This is based on highly counterintuitive 
data (Aronson and Carlsmith 1962; Carlsmith and Aronson 1963) that 
have been difficult to replicate (Sampson and Sibley 1965). (3) The 
surprise resulting from experiencing a discrepancy between expectations 
and obtained performance causes the individual to magnify the 
disparity. This is derived from contrast theory, attributed to Hovland 
and his colleagues (see also Anderson and Hair 1972 for additional 
discussion of this). (4) Later adaptations of contrast theory suggest 
that both assimilation and contrast are processes available to the 



129 
individual; the choice of which is engaged is dependent on 
characteristics of the situation (e.g., the degree of disparity between 
expectations and reality--Sherif and Hovland 1961). 

There are undoubtedly influences on perception beyond those 
outlined above, but in most cases these will be situation specific. As 
Lynch (1984a) and others have argued, it may be true that many of the 
perception effects obtained in behavioral experiments are artifactual 
and do not influence behavior. There is a trade-off between using more 
concrete attributes that allow quantification of effects and using more 
ambiguous attributes that may be more realistic in the generation of 
those effects (Zeithaml 1981). For the most part, then, perception in 
satisfaction studies is likely to have been fairly close to 
performance, at least as manipulated, and not too seriously biased by 
other factors. 

Finally, Woodruff, Cadotte and Jenkins (1983a) have recently 
elaborated on the perception construct and how it may function in the 
process that results in a satisfaction feeling. They contend that 
performance perception is always salient after an experience, but that 
any discrepancy with expectations may or may not be above a threshold 
of detectability . In effect, they argue that perception does not occur 
as a continuous mapping of performance attributes, but rather it 
results in a discontinuous series of summary evaluative "zones". The 
range that includes those levels of performance equivalent to that 
which was expected is labelled the "zone of indifference". Only- 
discrepancies beyond this range gain the attention of the individual. 
Thus, there may be some discrepancy threshold (or ambiguity threshold) 



130 
that determines whether the net perceptual process is to be data-driven 
or concept (expectation) -driven (Lachman, Lachman and Butterfield 1979; 
Neisser 1967) . 

The conceptualization presented herein takes a slightly different 
view of the process. In the current model, not all perceptions of 
performance reach high enough levels of attention to gain awareness. 
Similar to the Woodruff, Cadotte and Jenkins' scheme, however, 
experiencing extreme performance can attract attention. This, in turn, 
can result in a comparison against expectations and an appraisal of the 
commitment. In general however, a minimal level of response 
involvement (interest, attention) is necessary to justify the effort of 
engaging in the disconf irmation process (this involvement may be 
aroused through perception of extraordinary performance or other 
aspects of the situation) . 
Response Involvement 

As is true with many topics in marketing and consumer behavior, 
involvement is considered to be an important but not well-understood 
concept (that is, there is little agreement as to the treatment of its 
specific causes, effects and components). This has led to a number of 
inconsistent or contradictory theories, many of which are associated 
with some empirical validation attempts. By and large, conclusions 
drawn from this literature are no less equivocal than those based on 
the CS/D literature. 

Nevertheless, there is some consensus that the broad concept of 
involvement as a hypothetical construct is likely to influence 
cognitive processing (Burnkrant and Sawyer 1983). The involvement 



131 



concept has influenced thinking and research concerning the 
satisfaction construct and was useful in deriving the current 
conceptualization (Richins 1985; Richins and Bloch 1986, 1987). The 
definition and operationalization of the involvement construct that 
were chosen are somewhat more focused, but are not intended to be 
conclusive for all purposes in all fields. 

The general concept of involvement traces back to the earliest 
research in the CS/D literature. For instance, Cardozo's (1965) 
"effort" manipulation would be treated as involvement under some 
conceptualizations. Westbrook (1980a) measured effort and argued that 
it reflected different levels of involvement. This latter 
operationalization is closer to the manner in which involvement is 
conceptualized in the dissertation. That is, Cardozo's manipulation 
merely showed that subjects in each condition (high versus low search 
"effort") were equally willing to cooperate with the experimenter's 
definition of a necessary task. In the Westbrook study, greater effort 
could indicate greater interest in the product, greater attention to 
the decision or other such motivational states. 11 

The studies just discussed were concerned with the effects on 
satisfaction of varying the level of involvement. This question is 
logically preceded by an issue that has had little specific attention 
in the CS/D literature. That is, the question of whether "true" 



11 The possibility of these is acknowledged to suggest 
involvement. It is not argued that the construct was or was not 
operationalized in this particular investigation in a manner consistent 
with the perspective taken by the dissertation. 



32 



Table 6 
"Triggers" of Response Involvement 



1 . The item and/or the purchase occasion has some special 
significance for the individual. 

2. The social context in which the purchase is made and/or the item 
is consumed calls attention to the product and/or the purchaser. 

3. The consumer has had previous experiences with the product or 
service which suggest caution. 

4. The consumer has been advised to be careful in making the purchase 
by friends, consumer organizations, or consumer protection 
agencies . 

5. The consumer is inexperienced and poorly informed about the 
purchase and use of the product and is more conscious of all 
aspects of the situation than would normally be the case. 

6. The consumer encounters some unexpected circumstance which 
suggests caution, receives information or advice from a 
salesperson which is in conflict with prior beliefs, or otherwise 
encounters a "surprise" in the purchase situation. 

7. The consumer discovers after purchase that the product does not 
have all of the expected features, fails to perform as expected, 
or is defective or flawed in some way. 

8. The consumer discovers after purchase that the product has 
desirable qualities or features which were not expected or 
otherwise performs at a higher level than expected. 



;From Day 1977) 



33 



satisfaction can be evoked in situations of "true" low involvement is 
still outstanding (e.g., Zimmer and Deshpande 1984). 

Day's (1977) discussion of cues that may trigger the evaluative 
process that results in satisfaction provides the foundation for the 
present explication of involvement. The examples he provided of such 
cues (Table 6) parallel the representation of response involvement 
provided herein (Figure 9). That is, a high level of involvement may 
arise early in the purchase process as a result of prior knowledge 
(Day's examples 1-3), or very late, as a result of something perceived 
about the product (Day's examples 6-8), or at some point in between 
(Day's examples 4-5). The major contention, however, is that some 
"critical mass" of involvement must be attained in order to initiate 
the evaluation and evoke emotion (see also Olson and Dover 1976). 

The question of whether or not satisfaction always results from a 
consumption experience has been discussed at length. It has also been 
suggested that the more extreme behaviors thought to follow 
satisfaction require the higher motivation level that is generated by 
the emotion (compared to that derived from simple attitude). 
Involvement is conceived of as a kind of activation or arousal which 
enters the process and becomes the source of motivation (analogous to a 
conservation of energy principle) (Bloch and Richins 1983). Thus, to 
generate the higher motivation necessary for emotions, the involvement 
input would need to be greater than what might be required for attitude 
formation (or some other, less strong response) (see Locke 1969 for a 
similar model). The involvement literature itself generally deals with 
communication situations and correspondingly low involvement and 



34 



motivation levels, so it is not clear how a complete model of the 
interaction of these concepts may be developed. 

Involvement has been defined in a variety of ways. Rothschild's 
concept is close to that just explicated, "To me, involvement is a 
state of interest, motivation or arousal; in turn effort is a function 
of the level of involvement" (1984, p. 216). Ihis is more motivational 
than the result-oriented definition that he had at one time presented 
for response involvement , "the complexity or extensiveness of cognitive 
and behavioral processes characterizing the overall consumer decision 
process" (Houston and Rothschild 1978b, p. 185; see also Stone 1984, for 
a similar discussion) . 

Several authors have attempted to separate the issues of 
importance and involvement (Bloch and Richins 1983; Cohen 1983). Cohen 
(1982) also wrote of involvement as an "activation level" which may or 
may not be directed at a particular target. Operationalization of the 
construct "requires some measure of focused attention, ideally one 
which takes account of capacity available at the time for competing 
tasks" (p. 3; also, Greenwald and Leavitt 1984). 

All of these conceptualizations are generally consistent with a 
position that resulted from a special conference on involvement. "The 
proposed generic definition was that involvement is a state of 
motivation, arousal or interest. This state exists in a process. It 
is driven by current external variables (the situation; the product; 
the communication) and past internal variables (enduring; ego; central 
values). Its consequences are types of searching, processing and 
decision making" (Rothschild 1984, p. 217). Scales by which to measure 



135 



involvement have been developed (Belk 1984a; Bloch 1981; Sherrell and 
Shimp 1982) but care must be taken that related but undesired 
constructs are not confounded in the measurement (Cohen 1982). Most 
of the recently developed scales deal more with enduring than 
situational involvement (e.g., McQuarrie and Munson 1987, undated; 
Zaichkowsky 1985) . 

In adopting the generic definition for the current 
conceptualization, some side issues arise. As discussed previously, 
the present use of the decision construct does not imply a high level 
of commitment (as is usually thought to be synonymous with arousal of 
involvement). Rather, the model posits some change in cognitive 
processes (including those relating to the self--hence, ego 
involvement) that results from making an irrevocable selection. There 
is some evidence from the CS/D literature (Muncy and Hunt 1984; Olson 
and Dover 1976; Winter 1974) and other satisfaction literature 
(Freedman 1964; Locke 1967) that decision effects are discriminable 
from the effects of variables which should function through 
involvement . 

One such variable that is commonly included but not analyzed in 
consumer behavior studies is ownership (or personal relevance--Petty 
and Cacioppo 1986). A primary difference between most laboratory 
experiments and most field surveys is that ownership is seldom 
transferred in the laboratory. While several researchers have noted 
this discrepancy and included change of ownership as part of a study, 
ostensibly to increase involvement (Anderson and Hair 1972; LaTour and 
Feat 1980; Petty, Cacioppo and Schumann 1983; Winter 1974), they failed 



36 



to report whether or not the strategy succeeded. It may likely be the 
case that being given an inexpensive object is not sufficient, in 
itself, to evoke "motivation, arousal or interest". Thus, the question 
of whether ownership affects involvement and, in turn, arouses 
satisfaction has not been answered. 

Oliver and Bearden (1983) attempted to integrate involvement into 
satisfaction concepts and models. Although the issue they raised is 
clear, flaws in the methodology of the study render their results 
uninterpretable . For example, the cover story used involved "free 
sampling" so no commitment was generated. Second, and more important 
for the present discussion, the special nature of the product (an 
amphetamine diet aid) suggests that all users were likely to be highly 
involved. Even the authors acknowledged that the major "manipulation" 
(actually, a discrimination based on a rating scale) may not have been 
effective . 

Whereas Oliver and Bearden may not have measured low enough levels 
of involvement to discern the effects of various levels of the 
construct, other studies reported in the CS/D literature may not have 
generated sufficiently high levels to initiate the satisfaction process 
at all. The dissertation is concerned only with this latter issue, 
determining whether sufficient involvement is a prerequisite to evoking 
the satisfaction feeling. Later research might address the effects of 
intermediate levels of involvement. 

Of the studies for which the level of involvement reached is 
suspect, there are two types. In some studies, ownership is 
transferred but the product values are so low and nothing else about 



n; 



the situation is remarkable (see Table 6) that at least some 
respondents were probably best described as "unemotional" (e.g., 
Anderson and Hair 1972; LaBarbera and Mazursky 1983). The clearer 
deficiency, of course, involves cases in which respondents were never 
led to believe they would own the "target" object (e.g., Olson and 
Dover 1976; Surprenant and Churchill 1984; Wilton and Tse 1983). 

In summary, the construct adopted within the current 
conceptualization is similar to the definition of involvement that was 
proposed by the conferees at New York University and the 
ope rationalization of involvement proposed by Houston and Rothschild 
(1978a, b). That is, response involvement may be aroused by enduring 
and/or situational factors; and ownership is likely to be a 
contributing but not sufficient condition for such arousal. Finally, 
the entire discussion of involvement is tied to Day's (1977) suggestion 
that satisfaction can only result when involvement was high enough to 
drive the evaluative process, conf irmation/disconf irmation. 
Conf irmat ion/Pi sconfirmat ion 

As depicted in Figure 9, each of the preceding factors contributes 
directly or indirectly to the conf irmation/disconf irmation process or 
judgment (for simplicity, this is referred to hereafter as 
disconf irmation) . Although this concept lies at the heart of the 
disconf irmation paradigm, it has received comparatively little direct 
attention. In general, some type of valenced disconf irmation (Aronson 
and Carlsmith 1962) is included in a theory axiomatically and attention 
is focused on other constructs. However, this construct may represent 
either a comparison process that results in satisfaction or a judgment 



138 
that has been made en the basis of such a comparison (Oliver 1980a; 
Woodruff, Cadotte and Jenkins 1983a). Within the dissertation, 
disconf irmation is treated in the former sense, as a process. Although 
a judgment may also be made, it may not always be a conscious or 
salient result. Thus, the cognitive process is viewed as necessary for 
evoking the satisfaction feeling, but a specific measurable cognitive 
result of such a process is not required (more like Anderson's 1973 
conceptualization). The argument here is similar to that raised by 
Zajonc (1984) that the emotion is more accessible than the cognition. 
At this point in the development of the present conceptualization, 
however, the details of this process are not proposed. Rather, the 
concept will be used at its more general level, consistent with most 
treatments . 

As discussed previously, the disconf irmation paradigm, and 
therefore some form of the disconf irmation construct has been integral 
to the development of various satisfaction literatures. The early CS/D 
papers (Anderson 1973; Cardozo 1965; Olshavsky and Miller 1972), 
concerned with product evaluation, made a strong case for 
disconf irmation as an intervening process in quality assessment. These 
theories were based on perceptual theories that were discussed in a 
previous section. The use of product evaluation as the dependent 
variable suggests that the researchers believed that it occurred at the 
end of the process under investigation. This assumption requires a 
feedback process, however, inasmuch as disconf irmation cannot occur 
without some form of perception. These early theories thus required 



39 



that assimilation-contrast (Sherif and Hovland 1961) or other such 
processes both depend upon, and influence, perception (Filer 1952). 

As interest shifted to the more emotional satisfaction result, 
less was said about disconf irmation influences on perception. Although 
expectations are still hypothesized to have a potential effect on 
perception, disconf irmation is thought to result from the combined 
influence of these concepts. Furthermore, as mentioned in previous 
sections, the expectations are tempered by the decision that was made 
(Cialdini et al. 1978) and perception is affected by the experience in 
which it is gained (e.g., personal experience versus second-hand 
information) . 

These formulations of the disconf irmation paradigm seem to assume 
implicitly or explicitly (Woodruff, Cadotte and Jenkins 1983a) that the 
cognitive process always occurs (Oliver 1980b,c). As discussed in the 
previous section it is not clear that this is true, especially when 
nothing has occurred to arouse higher levels of involvement (Day 1977). 
The difficulty in demonstrating this phenomenon arises from the 
tendency for measures of disconf irmation to trigger the process, much 
as attitude measures are believed to prompt attitude formation (Fazio 
and Zanna 1981). Even if the process does not involve as much 
"cognitive algebra" as some propose (Oliver 1980b), some motivation is 
likely to be required for the process to occur. 

Oliver is among the researchers who treat disconf irmation models 
at attribute levels. In contrast, Day (1983) and the present 
conceptualization view expectations, perception and disconf irmation at 
summary levels. The difficulty with the former viewpoint lies in 



140 
operationalizing salient and important attributes at all phases of the 
process, even in light of the "unforeseen" triggers of disconf irmation 
(Table 6). 

The problem of determining at what level the process occurs 
becomes magnified when integrated into the larger satisfaction models. 
For example, attempts to tie disconf irmation to repurchase intention 
(Oliver and Linda 1981) do not yield strong results when specific 
attributes or even summed attribute scales are used. Similarly, 
theorizing about the result of simple confirmation (Swan and Trawick 
1979) depends on the salience and importance of the confirmation 
judgment. More recently, Oliver and Bearden (1985) report evidence 
that overall disconf irmation was a mediator between "inferred" 
disconf irmations and overall satisfaction. 

The problems in the definition and operationalization of the 
construct are also reflected in its measurement (Shuv-Ami 1984). Early 
research simply defined disconf irmation as the difference in value 
between expectations and perceptions. Later work demonstrated that 
rating scales (better/worse than expected) are improved methods (Moore 
and Shuptrine 1984; Prakash and Lounsbury 1983), but subject to the 
experimental reactivity problem discussed previously. Selection of a 
measurement technique, then, will depend on the purpose of the 
investigation and the sensitivity of any theory involved to possible 
contamination. 

In general, the recent use of the disconf irmation construct has 
been to assess its effect on satisfaction (e.g., Swan 1987). In such 
cases, rating scales may sensitize subjects to the extent of 



141 
disconf irmation. On the other hand, few investigators have 
demonstrated concern over initiating satisfaction processes by 
experimental situations or measurement instruments. 

Most such investigations have shown that disconf irmation does 
significantly affect satisfaction (e.g., Kennedy and Thirkell 1987; 
Moore and Shuptrine 1984). Moreover, this influence is directional; 
positive disconf irmation prompts satisfaction and negative prompts 
dissatisfaction, in contrast to the early models that posited 
dissatisfaction resulting from any disconf irmation (Aronson and 
Carlsmith 1962; Carlsmith and Aronson 1963; Sampson and Sibley 1965). 

Few of these models, however, have explained how satisfaction 
results from a cognitive process. In the discussion of emotions above, 
it was stated that emotions result from cognitive processes--even if 
such processes are not conscious. Thus, by defining satisfaction in 
emotional terms and defining disconf irmation as a cognitive process 
rather than a cognitive result, the present conceptualization 
integrates emotional and satisfaction theories in a manner that does 
not conflict with recent debates as to the precedence of cognitions and 
affect (Cohen 1981; Fazio, Powell and Herr 1983; Zajonc 1968, 1930). 

In summary, the conceptualization of satisfaction that is being 
proposed treats disconf irmation as a necessary result of the presence 
of the other antecedent conditions that have been discussed. 
Moreover, some level of satisfaction will result from this comparison 
process— with no other construct currently hypothesized to intervene or 
moderate. The disconf irmation process operates at a summary or overall 
utility level, although it may be triggered by a particular salient 



142 
attribute. Satisfaction will not result in the absence of the 
disconf irmation process (nor, therefore, in the absence of any of the 
other antecedent conditions). 

Purpose of the Extended Conceptualization 

As presented, the current "six-antecedent" model may appear to be 
an unnecessary complication of the more commonly accepted 
"three-antecedent" formulations. It is useful, at this point, to 
reiterate and clarify the purpose of the conceptualization. 

Prior research has sought to understand levels of satisfaction; 
i.e., what causes a person to experience degrees of the feeling and how 
subsequent behavior is affected, in turn, by the feeling. The three 
commonly implicated constructs, expectations, perception and 
disconf irmation have been relatively consistent significant influences 
on measured satisfaction, albeit not always accounting for a great deal 
of variance. 

A potential cause of this loss of explanatory power has been 
uncertain definitions and operationalizations of the key dependent 
construct. That is, the situations contrived by researchers have not 
been demonstrated to replicate the conditions normally encountered in 
the marketplace. If interest in satisfaction arises from the desire to 
understand a market phenomenon, this would seem to be a serious flaw. 
The problem is exacerbated by the inability of current measurement 
techniques to assess the ability of an experimental situation to evoke 
satisfaction. 

Thus far, the dissertation has shown how the satisfaction 
construct relates to other purchase and post-purchase phenomena (e.g., 



'4 3 



attitude, complaining) and has based a definition on this theoretical 
network. It has also proposed that the pattern of emotions that 
describes the dissatisfaction — satisfaction continuum can be measured 
with sufficient sensitivity to reveal the absence of the feeling. 

The model of the satisfaction process presented herein (Figure 9) 
takes an additional step. In order that an emotional response that can 
be called satisfaction will result, a situation must encompass all of 
the antecedent conditions presented in the model. This 
conceptualization, then, provides a template with which one may 
identify, describe, or create a satisfaction-evoking situation. 

The conceptualization might at first appear to comprise a rather 
restrictive set of criteria. It is important, however, to note that 
the antecedent conditions are not difficult to identify in the 
marketplace. In fact, the three concepts added by the dissertation, are 
the net result of a comparison of "typical" marketing conditions that 
result in satisfaction with conditions generated ostensibly to study 
the construct . ' 2 

For example, it has been argued that the decision construct can be 
represented by the purchase act. Satisfaction surveys and reports of 
complaining generally deal with items purchased by the 
respondent/complainer (or a decision making unit which included the 
respondent). These surveys appear to assume that a purchase 
encompasses the characteristics of volition, selection and 
irrevocability (in that the transaction has been consummated and 



12 Again, the conceptualization generalizes to all satisfaction 
domains. The example of consumer behavior in the marketplace is used 
for consistency and clarity. 



144 
replacement with a different product entails some effort and 
"complaining") that were deemed necessary to constitute decision (e.g., 
Andreasen and Best 1977). 

Similarly, marketers interested in assessing levels of 
satisfaction with their products direct their surveys to those who have 
had personal experience with the item in question. A mail survey by a 
major domestic automobile manufacturer carries this instruction in bold 
print, "Please have the person who drives this car the most fill out 
this questionnaire". It is also common practice in the conduct of 
multi-product surveys (i.e., conducted in attempt to compare 
satisfaction levels across product classes) to not collect satisfaction 
ratings from respondents who report that they rarely or never use items 
in a particular category (e.g., Day and Landon 1976). 

Comparison studies such as these often identify more expensive 
products as the ones most responsible for consumer satisfaction and 
complaining (e.g., Shuptrine and Wenglorz 1930, 1981; TARP 1979). 
Although the increase in complaining activity may be partially 
attributable to the higher return expected from complaining about more 
costly items, this explanation does not hold for increased mention in 
satisfaction surveys. Quite simply, satisfaction responses to "high 
involvement" products (expensive, owned by the respondent) are 
evidently more salient and/or memorable. Other involvement arousing 
factors (total failure, physical danger) are also frequently stated in 
solicited and unsolicited accounts of dissatisfaction (e.g., Best 
1981), although the activation level soon tapers off (Richins and Bloch 
1986). 



145 
The discussion in this section is not meant to provide conclusive 
evidence of the usefulness of the expanded framework. Rather, it 
suggests that the specified necessary antecedent conditions for 
satisfaction to result are not particularly stringent, given 
experiences from non-experimental settings. 13 
Adequacy of Prior Studies 

Given that the present conceptualization has been proposed as a 
template for the design of satisfaction research, it should also be 
useful in evaluating prior research efforts. This section of the 
chapter examines individual papers from the CS/D literature in light of 
the extended model and discusses the necessity of considering the 
entire situation in the design of satisfaction studies. Potential 
improvements to the studies that might have accrued from attention to 
the conceptualization are noted where appropriate. 

Table 7 summarizes this analysis (the structure is based on that 
used bv Wilkie and Pessemier 1973). The articles reviewed were chosen 
on several criteria: (1) articles considered to be "classic" in the 
CS/D literature (including those originally intended only to examine 
product perception), (2) articles that were noticeably deficient in one 
or more of the antecedent conditions, and (3) satisfaction articles 
that have appeared in a major journal ( Journal of Marketing ; Journal 
of Marketing Research; Journal of Consumer Research) or conference 



13 The complete model, rather than just the individual concepts 
outlined above, is supported by as many as 129 out of 138 descriptions 
of "satisfying" or "dissatisfying" occasions collected in pilot studies 
(similar to the critical incidents described by Maddox 1981 and Swan 
and Combs 1 976) . 



146 

Table 7 



Prior CS/D Research and 
the Extended Conceptualization 





a) Survey designs 










Citations 








Swan & 


Anderson 








Scott 4 


Bearden 


West- 




Combs 


et al. 


Oliver 


Westbrook 


Maddox 


Yalch 


4 Teel 


brook 


Issues 


1976 


1979 


1980a 


1980a 


1981 


1980 


1983 


1983 


Satisfaction Concept 
















Type* 


I 


I 


UO 


I,M,L 


I 


I 


S 


I 


Time b 


E 


E 


E 


E 


E 


A 


E 


S 


Definition 


C 


C 


A 


C,A 


C 


none 


none 


A 


Measures'* 


V,C 


V,C,B 


V E* 


V,C*,E* 


~ 




V,E 


V,G,C 
E*,B* 


Conditions* 


















Expectation 


? 9 


P 


P,S,C 


P 


P 9 


?,S 


P,S,C 


P 


Decision 


p? 


? 


P 


P 


P? 


?,M 


p 


p 


Personal 


















Experience 


p 


P 


P 


P 


P 


P 


P 


P 


Perception 


? 


P 


P,S 


P 


P 


P,M 


P 


p 


Response 


















Involvement 


? 9 


P 


P 


P_ 


? 7 


P,M 


?? 


? 


Disconf irma- 


















tion 


p? 


? 


P,S,C 


? 


P? 


P,M 


P,S,C 


P? 


Role of 


















satisfaction 5 


D 


D 


D,S 


D 


s 


" 


D,S 


D 


Satisfaction 


















outcomes? 


~~ 


-~ 


A,R 


" 




A,L 


CR,W 
L,AR 




Methodology notes 


















Subjects* 1 


S 


R 


R 


S 


c 


R 


R 


C 


Target clothing 


auto 


vaccine 


car, varied 


diet 


auto 


auto 



shoes cola 



Interpretation 

Design ' s 
conformance 
with concep- 
tualization 1 

Result ' s 
consistency 
with concep- 
tualization 



147 



Moore Richins 

4 Robertson 4 West- 

Shuptrine et al. Bloch Oliver brook 

1984 1985 1986 1987 1987 



s 


I, 


PO 


I 




s 




I 


E 


E 




E,A 




E 




E 


C 


c 




none 




A 




c 


V,C*,E* 


9 






V 


,C*,E* 


V 


, G , C* , 
E* 


P,S,C 


2 


s 


9 




P,S,C 




P,S,C 


P? 


P 




P 




P? 




P? 


? 


P 




P 




P 




P 


? 


9 




P 




P 




P 


P? 


P 




P , M , S 




P 




P 7 


P,S,C 


9 




■j 




P,S,C 




P,S,C 



D,S D ? D,S D,S 



AR,LR — W,L WR.LR CR 

CR 

R R R R R 

chari- Christmas cars MBA car, 

ties gifts program CATV 



148 



Table 7--continued 



b) Experimental designs 



Citations 



Cohen & Olshavsky Lambert LaTour & 

Cardozo Goldberg & Miller Anderson Winter et al. Oliver Peat 
1 965 1970 1972 1973 1974 1977 1977 1980 



Satisfaction Concept 


PO 

ne 
C 


I 

A 
none 


I 
A 
none 


I 

A 

C 


I 
E 
C 


S 
A 
none 


I 
A 
none 




Type 4 

Tirae b 

Def inition c 

Measures' 1 


I 
A 

nc 

V 


I 

A 
C 

V,C 


Conditions* 
Expectation 
Decision 
Personal 


A 


- 


P 
P 


P,M 
A 


?,M 
A 


P.M 
P 


p 
? 


P , M 
A 


?,M,C 
A 


experience 
Perception 
ResDonse 


p 




P 
P,M 


? 
P,M 


P 

P,D 


P 
? 


A 
A 


P,M 

?,D 


P 
P,D 


involvement 
Disconf irma- 


? 




? 


A 


P 


P 


A 


A 


? 


tion 


P 




?,M 


P,M 


P 


? 


P,M 


P.M 


P,M 



Role of 

satisfaction f D 



Satisfaction 
outcomes^ 

Methodology notes 
Subjects" S 
Target D e 



Interpretation 

Design' s 
conformance 
with concep- 
tualization 1 

Result's 
consistency 
with concep- 
tualization; 



L,A 



S 

coffee 



L,A 



L,A 



S S R R S R 
tape pen soap vendor car tile 
recorder pad cleaner 



i.A.9 



LaBarbera Mowen Wilton Surprenant Jordan Fisk Hoch Bolfing 

4 4 & & 4 & & Cadotte 4 

Mazursky Grove- Tse Churchill Leigh Folkes Young Ha et al. Woodruff Carsky 

1983 1983 1983 1984 1984 1984b 1985 1986 1987 1987 1987' 



I 


PO 


I 


I 


I 


I 


UO 


I 


S 




C 


S 


E 


A 


A 


A 


E 


E 


A 


A 


A 




E 


E 


C? 


c? 


C 


none 


none 


none 


C 


-- 


C 




C 


C 


V,C 


v,c 


V,C 


V,C 


V,E 


— 


7 


" 


V 


E 


? 


V,E 


P? 


? 


-P,M,S 


P,M 


P^ 


? ? 


? 


P 


? 


M 


? 


?,s 


P 


? 


A 


A 


A 


A 


A 


A 


? 




A 


A 


P 


A 


P 


P,M 


A 


A 


A 


P 


p 




P 


A 


? 


A 


P,M,S 


P 


7 


P 


?,M 


-, 


? 




i 


?,S 


P? 


A 


A 


7 


A 


A 


A 


A 


? 




P,M 


A 


?? 


P? 


P,M,S 


P,M 


A 


P 


P,M 


? 


1 







■> 



D,S D D,S D,S D,S D — D D D,S 

L,R W,L W,C,A,LR L L W,C,L L ~ — — W,L,A 

R RS R RSS.RSRS S 
grocery compact plant video- airline shirts/ rest- 
products auto disk game varied service towels aurant wine cars 



150 



Notes: a: P0 = purchase occasion, U0 = use occasion, I = item, C = 

class, S = store or vendor, 
b: A = acute, E = enduring, 
c: C = cognitive definition, A = affective definition, none 

= no explicit concept, 
d: V = verbal, G = graphic, C = cognitive, E = emotional, 8- 

= behavioral, * = reliability test reported, 
e: P = present, A = absent, ? = cannot determine, M = 

manipulated, S = scaled as intermediate variable, C = 

causal relationship with satisfaction. 
f: D = dependent, S = scaled as intermediate variable, 
g: C = complaints/compliments, W = word-of-mouth, L = 

loyalty/ exit , A = attitude, R = resultant from 

satisfaction, 
h: S = students, C = convenience, not strictly student, S = 

representative (random) sample attempted. 
i: Y = yes, study design conforms to criteria of expanded 

conceptualization, N = no, study design does not 

conform, 
j: S = support for conceptualization implied by results, ? 

= results ambiguous with respect to conceptualization. 



151 
proceedings (Advances in Consumer Research and CS/D&CB conference 
papers are examined only as selected by the other criteria) . 

The table is divided into two parts to facilitate presentation and 
to reflect slightly different research goals and traditions. Reports 
of experimental investigations are summarized in part "a" of the table; 
surveys are listed in part "b". The discussion will focus on the 
experiments since correction of those flaws is a major purpose of the 
dissertation. The surveys are of interest because they reflect the 
constructs in realistic settings. In general, the surveys are not 
sensitive to conceptual differences between satisfaction and attitude, 
which opens some avenues to improvement via the extended 
conceptualization. 

The same issues are summarized for each set of studies. The rows 
of the table identify attention to, incorporation of, or a position 
taken on particular conceptual or methdological points. "Type", for 
example, refers to the satisfaction types suggested in Table 2. 
Satisfaction emotions can be aroused by targets at various levels of 
abstraction. Although the process is thought to be essentially the 
same, specific precursors and strength of emotions are likely to vary 
across the hierarchy. 

The next row, "Time", indicates the interval addressed by the 
investigation. The dissertation defines satisfaction to be emotional 
in nature; and quick to decay. This is more accurately reflected in 
the concept of acute rather than enduring satisfaction. The latter 
type, however, seems to have been of more interest to policymakers and 
other practitioners interested in long-run market reactions. 



152 
"Definition" attempts to identify the conceptual treatment of 
satisfaction. In some cases where no explicit definition was presented 
in the paper, it was possible to characterize the position taken based 
on either the context of the article or prior reports by the authorfs) 
(e.g., Oliver 1980a, 1987). The "Measures" used to assess satisfaction 
are characterized based on the categories in Figure 4. If a 
reliability test is reported, this is noted. 

Most crucial to the contribution of the dissertation is the 
examination of the critical "Conditions." In some cases, it is evident 
from the description of the methodology that a particular element of 
the process was (or was not) within the respondent's experience. Many 
of the constructs were explicitly manipulated or scaled as part of the 
investigation. In many instances, it is necessary to infer from the 
description of an experiment or from the qualifications of a survey 
sample whether or not the construct was present. If some, but not all, 
subjects experienced the enabling condition, it is generally identified 
as present. Finally, some studies sought causal relationships among 
variables. If evidence was found for a causal relationship with 
satisfaction, this is noted. 

Not all of the papers summarized in Table 7 sought to investigate 
satisfaction—some never invoke the construct. Among these are papers 
that dealt with dissonance (Cohen and Goldberg 1970), involvement 
(Richins and Bloch 1986) and attitude (e.g., Winter 1974). Others 
scale satisfaction as an intermediate variable between pre-purchase and 
post-purchase constructs or outcomes (e.g., complaining, word-of- 
mouth) . If a test of causality showed that a state or behavior 



53 



resulted from satisfaction, this is indicated. Inclusion criteria for 
such studies were broadly defined since it was felt that the proposed 
conceptualization makes recommendations valid for a variety of post- 
purchase investigations. 

Table 7 notes other methodological points that are often of 
interest in summarizing a body of literature, the subject pool, and 
target stimuli used. Student and other convenience samples are 
identified discretely from sampling plans that attempted some degree 
of representativeness--there were few truly random samples. For one 
study (Oliver 1987), a student sample was particularly appropriate for 
the issue under discussion and was categorized as representative. The 
"Target" identified is the stimulus used or manipulated. It is not 
always clear that subjects responded at this level, some seem to be 
willing to attribute item problems to store causes, or the reverse, and 
complicate measurement. 

The interpretation portion of Table 7 attempts to summarize the 
implications of each study in light of the proposed conceptualization 
and to evaluate the conceptualization against the empirical evidence in 
the study. The summary entries represent post hoc evaluations that 
allow considerable margin for bias. Thus, while none of the studies 
was judged to discount the conceptualization, it may be due to a broad 
concept of support. This notion of support, however, is secondary to 
the purpose of the table, which is to show how the research may have 
been better by attending to necessary factors in the designs. 



154 



Conclusions from the literature summary 

Table 7 is meant to be neither an exhaustive nor representative 
random list of CS/D studies. Nevertheless, it is instructive to look 
at patterns which may be found across the rows of the table. In 
particular, some of these patterns illustrate differences between 
experimental and survey research. 

Type . For the most part, both series of studies have concentrated 
on item-specific satisfaction. This is not surprising and consistent 
with much of research in marketing and consumer behavior. Following 
attitude research tradition, CS/D researchers have found that 
individual items have more manageable attribute profiles than stores, 
product classes, and other more general consumer targets. On the other 
hand, reactions to experiences as specific as purchase or use occasions 
may be too varied to be predictable or interpretable . That is, 
consumers may view one instance of unusual performance as a rare 
event" rather than as a disconf irming experience. 

Time . The amount of evidence, and impact of disconf irmation, will 
vary across the time dimension. It is evident from Table 7 that survey 
and experimental designs tap different portions of the time continuum. 
Most experiments are conducted in a single session, allowing no time 
for satisfaction to erode. Surveys, on the other hand, usually measure 
response after relatively longer usage--owing to the difficulty of 
identifying and locating a sufficient number of respondents in an acute 
satisfaction stage. There are some exceptions in each case. Some of 
the experiments have utilized more longitudinal designs and have 
measured more delayed responses. Surveys have used critical incident 



155 



techniques to assess response at an earlier, more acute, time or have 
argued that very recent purchasers of highly involving products still 
experience stronger emotional response (e.g., Richins and Bloch 1986). 
This variation of time levels makes comparison of experiments and 
surveys untenable. It has been argued previously that satisfaction 
soon decays into attitude, which may be the construct measured by most 
of the surveys. If this is the case, then the findings of relatively 
low rates of satisfaction or dissatisfaction related behaviors in the 
marketplace are more plausible. Marketing practitioners should not 
expect complaining or word-of-mouth rates to correlate with levels of 
post-purchase attitude. 

Definition . The conceptual distinction between satisfaction and 
attitude is best communicated by well-articulated definitions. This 
seldom has been the case in the studies reported. Many of the studies 
fail to provide an explicit definition of satisfaction, due either to a 
presumption that the concept is clear from prior research or to sheer 
neglect. Table 7 reports the types of definition provided (or, 
implicit in the treatment of the construct) or an indication that none 
existed. For those studies that did not purport to measure 
satisfaction, it is reasonable that the construct was not defined. 
What is more alarming is the frequency of investigations, often 
experiments, that neither define the construct nor take an identifiable 
position on its nature. In some cases, a report of the measurement 
scale(s) used for satisfaction suggest the philosophy adopted, but too 
often there are inconsistencies in the set of measures adopted or 
between measures and an espoused concept. 



156 
Measures . Virtually all of the studies reported have gathered data 
using written questionnaires. The production mechanics involved have 
led to a dominance of verbal scales (i.e., it is much simpler to 
generate polished looking text than graphics). Within these, there has 
been a tendency to use scales that were more cognitive than emotional 
in nature. The experiments reported have been prone to this more so 
than the surveys. 

This measurement pattern across the literature creates an 
interesting discrepancy. The survey designs, which logically should 
be measuring something akin to attitude (enduring satisfaction), use 
more of the affective measures that should be sensitive to the 
emotions aroused by the particular item or incident (acute 
satisfaction). Thus some weaknesses in results may be reflecting these 
measurement errors (that can be cast in construct validity terms). 

Conditions . Discussing and measuring satisfaction in the context 
of a given investigation requires confidence that the construct exists. 
The dissertation's claim that this has not always been so is reflected 
in this portion of the table. Even allowing for broad presumptions of 
presence, not all of the enabling conditions fare well. Since the 
earliest days of the disconf irmation paradigm, the concepts of 
expectations, performance perception, and confirmation or 
disconf irmation have dominated post-purchase research. Even when 
neither manipulated nor measured explicitly, each construct seems to be 
an implicit part of nearly all of the studies. For the sake of 
clarity, however, it would have been better to explain explicitly one's 
underlying model and the role of each construct in it. This is 



157 
particularly true in the case of the survey designs which often seem to 
result from an assumption that the respondent has performed a full 
decision, purchase, and post-purchase task. Surveys of satisfaction, 
unaccompanied by measures of expectations or absolute performance 
perception, tell an incomplete story about what underlies satisfaction 
levels. 

More critical are the patterns described for the new constructs. 
For those, the experimental design portion of Table 7 is the more 
illuminating. A majority of the experimental studies showed no 
evidence of the subjects' being forced--or even allowed--to make a 
decision. Thus, the expectations were never acted upon in anything 
resembling a purchase task. 

The absence of decision may have occurred either in a setting 
that included personal experience or in situations in which 
performance perception was available only as a description. It is 
unfortunate that these role-playing studies have come to be so common 
in CS/D research. Again, the problem is much more severe in the 
experimental rather than survey designs. 

The trade-off being made is, apparently, an attempt to generate 
involvement or enhance other manipulations at the cost of realism. The 
argument is that role-playing enables the use of expensive or difficult 
to manipulate products (Fisk and Young 1985, Surprenant and Churchill 
1984), controls for past events, and prevents bias caused by deception 
(e.g., Fisk and Young 1985). While the concern over bias is admirable, 
the potential for bias seems much higher when both manipulations and 
measures are obtrusive (Sechrest 1979). 



158 
The desire to use expensive products as experimental stimuli often 
arises from a desire to create a more involving experimental situation. 
This would seem desirable given the conceptualization under discussion. 
Richins and Bloch (1986), among others, argue that this effort is 
misguided if the appropriate interaction of situation and enduring 
involvements is not taken into account. Indeed, the present chapter 
has argued that it is the personal relevance or "stake in the outcome" 
of the decision that is the important motivating force in the 
involvement mechanism. Thus, any benefit gained by using an expensive 
or interesting product is lost in the abstraction of role-playing. The 
experiments in Table 7 represent both entire role-playing situations — 
in which the manipulations are wholly hypothetical (e.g., Carsky 1987) 
— or partial role-playing — in which subjects are asked to imagine 
themselves to be buyers of a product that is physically present (e.g., 
Wilton and Tse 1983). In neither case, however, does the situation 
have any implications for the subject once the experimental session has 
concluded. These are noted as absent, or at best questionable, 
response involvement conditions. 

Satisfaction role and outcomes . Because of the selection criteria 
used, it is not surprising that some form of satisfaction was a 
dependent variable in many of the studies summarized. Inclusion of 
satisfaction as a scaled intermediary between pre-purchase states and 
post-purchase behaviors was also common among both surveys and 
experiments. This provides some evidence in support of the grander 
models of post-purchase behavior. 



159 
The selection of behavior is not the same as would have been 
recommended by the present conceptualization. Both surveys and 
experiments emphasized "loyalty" behaviors--purchase of same or similar 
item — as outcomes of interest. While these are likely to be important 
from a managerial standpoint, it has been argued that such a response 
is more appropriate to attitude — following the Theory of Reasoned 
Action — than satisfaction processes. In the experiments attitude was 
often measured as a resultant construct, but with little causality 
testing. Such testing was most common in the broader based surveys 
which utilized regression techniques and a wider selection of dependent 
measures . 

Methodology notes . Across the studies, use of student subjects is 
widespread. Although this is desirable in one instance (Oliver 1987), 
as noted previously, the general pattern is bothersome. The 
combination of role-playing methodologies, lack of personal relevance 
and the use of student subjects has probably yielded a number of 
artifactual results. The products used as stimuli, while seldom 
inappropriate for a student sample, have not been targeted specifically 
to such limited samples. The variety of targets used has uncovered 
difficulties in making manipulations salient or generating sufficient 
involvement . 

Interpretation . Rather than attempt to globalize the 
interpretations at the bottom of the table, some comments can be made 
regarding selected papers as viewed from the vantage point of the 
extended conceptualization. The earliest papers generally thought to 
be satisfaction experiments (Anderson 1973, Cardozo 1965, Olshavsky and 



160 
Miller 1972) dealt more with product perception and consistently lacked 
any opportunity for the subject to make a decision regarding the 
product. This omission may help to explain why the researchers failed 
to find significant contrast effects: the expectations weren't "fixed" 
by an irrevocable commitment. Cardozo (1965) alluded to a contrast 
effect, but this study dealt more with equity of the experimental 
(shopping?) situation than with actual product involvement. That is, 
the expectation that was disconfirmed was the type and value of the 
gift received rather than its functionality. In general, these studies 
report findings consistent with simple perceptual or attitudinal 
processes . 

On the other hand, an early investigation that did follow the 
conceptualization well (Cohen and Goldberg 1970) is often slighted in 
the satisfaction literature. While these authors did not claim to be 
engaged in satisfaction research, the conditions they set forth to 
generate dissonance plus post-experience reevaluation fit the criteria 
displayed in Table 7. Although none of their measures was adequate for 
satisfaction, the loyalty and attitude measures behaved as expected. 
Thus, while the study does not contribute specifically to knowledge 
about the satisfaction construct, it is a useful model for the design 
of the satisfaction investigations. 

Similarly, the study reported by Winter (1974) fits the criteria 
but was positioned as a study of attitudes. The arguments offered to 
distinguish a trial purchase from an adoption purchase offer weight to 
the necessity of the decision construct. Winter reports that "the 
purchase act, even for a low-involvement product such as scouring pads, 



161 



may be instrumental in altering consumer cognitions in the post- 
purchase period before actual consumption of the product takes place 
(1974, p. 170). The satisfaction process extends this reevaluation 
beyond the dissonance stage. While the importance of the decision 
construct has been highlighted, the lack of satisfaction measures 
hinders a more general endorsement of the conceptualization. 

The Lambert et al. (1977) paper is notable for its attempt to 
evaluate an industrial purchase process. While laudable for this goal, 
the lack of actual decision and the role playing structure of the 
situation hinder any interpretation of the findings. This lack of 
decision also reduces confidence in the findings of other papers that 
are discussed in more detail elsewhere (LaTour and Peat 1980, Oliver 
1977). 

An investigation of "product trial" by Scott and Yalch (1980) is 
another study that could have been cast as a satisfaction study, but 
was not. As with the Cohen and Goldberg effort, this experiment 
fulfilled the criteria but included no measures of satisfaction nor 
discussion addressing directly the construct. Personal experience was 
found to affect evaluation of the target product as subjects gathered 
progressively more information about the product. 

A set of role playing experiments began the 1 980 ' s (Mowen and 
Grove 1983, Surprenant and Churchill 1984, Wilton and Tse 1983). The 
findings are consistent with cognitive operationalizations of 
satisfaction but can also be the result of experimental demands. Less 
obtrusive or emotional measures may have provided some divergence from 
intuition. Wilton and Tse (1983) reported some divergence between 



162 
attitude and satisfaction measures, colored somewhat by a surprisingly 
dispassionate assessment of a high-involving product. 

The Jordan and Leigh (1984) and Folkes (1984b) papers illustrate 
the dangers inherent in casual adoption of a paradigm. The former does 
not even match the "old" paradigm in that no disconf irmation is 
suggested. The authors merely assess post-purchase attitudes among 
children. Folkes, on the other hand, does not invoke the satisfaction 
construct or a satisfaction model but the examination of post- 
consumption attributions and behaviors certainly entails such emotions. 

Fisk and Young (1985) argue that role-playing was necessary for 
the selected product class (airline service), but never present a case 
for the appropriateness of the product class. It is impossible to 
infer the effects of the lack of decision, experience, and involvement 
on the criterion variables because no means were reported. The paper 
also did not discuss explicitly satisfaction results. In general, the 
paper reflects a casual interpretation of and approach to the 
disconf irmation paradigm. 

Among the most recent experiments reported, there is a distressing 
tendency to maintain the role-playing approach (Bolfing and Woodruff 
1987, Carsky 1987, Hoch and Ha 1986). The concepts and definitions of 
satisfaction remain cognitive, although there is more emotion or affect 

reflected in some of the measures (e.g., How do you feel about ?). 

Nevertheless, the ability of such situations to arouse true emotions 
remains suspect. Bolfing and Woodruff (1987) claim to increase 
involvement, but in the absence of any personal relevance. 



163 

Cadotte, Woodruff, and Jenkins (1987) executed a design that fit 
the criteria by including an actual purchase in a field experiment. 
Their interest in satisfaction with the store (in this case, a 
restaurant) makes the results somewhat more difficult to compare with 
product satisfaction studies, but the old and extended disconf irmation 
models are generally supported. In addition, the study extends the 
concept of comparison standard beyond what had been considered simple 
expectations (although consistent with the dissertation's viewpoint). 

The implications of the survey reports are easier to summarize; 
most of these addressed actual purchase, consumption, post-purchase 
phenomena. Thus, it is possible to infer— if not fully demonstrate — 
the presence of the enabling constructs in most survey accounts. 

Two studies sought to provide a new, Two-factor theory, approach 
to satisfaction (Maddox 1981, Swan and Combs 1976). The critical 
incident technique used in both of these studies is likely to result in 
descriptions of consumer experiences which fulfill the criteria, but 
those can only be interpreted on a case specific basis. Had the 
studies used satisfaction measures--particularly emotional ones--rather 
than simply classification as satisfying or dissatisfying experiences, 
the conceptualization may have garnered more support. 

The Oliver (1980a) survey does provide more emotional measures and 
attempts to distinguish satisfaction from attitude. The path analysis 
reflects a separation between the constructs such that satisfaction is 
found to be a significant cause of post-decision attitude. Although 
the subjects involved did undergo processes defined by the extended 
conceptualization, Oliver seems to force the satisfaction construct 



164 
upon them. That is, non-satisfaction is not allowed. This criticism 
is not unique in this study — few, if any, surveys disqualify 
respondents for absence of satisfaction and/or dissatisfaction. This 
is reflected in somewhat muddled treatments of confirmed expectations 
and their effects on consumers. 

Similar weaknesses make comparisons across studies difficult. The 
Anderson, Engledow and Becker (1979) study was more concerned with 
demonstrating LISREL than providing substantive knowledge in CS/D. 
Other recent surveys provide little control over time elapsed since the 
disconf irming experience and may be reporting attitude results (Bearden 
and Teel 1983, LaBarbera and Mazursky 1983, Moore and Shuptrine 1984). 

Westbrook's studies (1983, 1987) provide a decidedly emotional 
approach to satisfaction, but also assume that every consumer can be 
placed along the continuum at any time. It may better serve his needs 
to focus respondents' attention on critical incidents and measure 
emotional response to those. This would provide firmer evidence that a 
specific disconf irmation is involved and the target of the response. 
One could argue that the focus on a particular item (i.e., automobile) 
is sufficient; but the confirming versus disconf irming experiences may- 
be ambiguous. 

On a final note, the Richins and Bloch investigation (1986) was 
not a satisfaction study, nor did they collect satisfaction measures. 
It is included as a survey that explicitly mentions each of the 
enabling constructs and specifically investigates one of them, i.e., 
involvement. Of particular interest to CS/D and other consumer 
behavior researchers is the authors' caution regarding the difficulty 



165 



of manipulating situation involvement in the laboratory. One 
conclusion that might be drawn from their study is that involvement is 
inherently an individual characteristic and is better viewed as a 
blocking variable in a quasi-experiment rather than a manipulable 
factor. 

General conclusions . Table 7 has a relatively simple overall 
story to tell. The main point is that the CS/D literature can be made 
more coherent by specification of criteria for satisfaction studies. 
Much effort is wasted on attempting to integrate results from 
inherently discrepant studies and paradigms. This prescription is 
particularly appropriate to experimental investigations. 

As discussed, surveys are not without fault. Although actual 
purchase and consumption are usually involved, the elapsed time 
suggests a confounding of satisfaction and attitude. More distinct 
concepts, measures and methodologies are needed if satisfaction is to 
be treated as a unique construct. 

The issues in Table 7 provide points of comparison among studies 
and research traditions. Perhaps a taxonomy can evolve that would 
enable better mapping from the laboratory experiment to the field 
(market) survey. 

Summary and Conclusions 

This chapter has integrated the literature in CS/D (reviewed in 
Chapter 2) with related literatures. The satisfaction construct was 
redefined as emotional in nature and generalizable to domains other 
than consumer behavior. A related concept, attitude was shown to be 
discriminable in its nature and effects on behavior. A general model 



166 



of the situation that results in the pattern of emotions defined to 
constitute satisfaction was presented and shown to be consistent with 
common consumer situations. The extant CS/D literature was then 
evaluated with reference to the model and it was demonstrated that a 
lack of adherence to the conceptualization may account for many of the 
weak and/or inconsistent results that have been reported. 

In the next chapter, the theory embodied in the model (Figure 9) 
is extended to specific hypotheses concerning the impact on 
satisfaction studies of omitting necessary conditions. A methodology 
for testing these hypotheses is then presented in Chapter 5. 



CHAPTER 4 

RESEARCH HYPOTHESES 

Introduction 

Preceding chapters have reviewed the evidence currently available 

regarding the antecedents and consequences of satisfaction. This 

review was augmented with discussions of literatures related to 

emotions and other operationalizations of satisfaction. The review 

culminated in the definition of satisfaction as a pattern of emotions 

which is evoked by a process slightly more complex than that assumed by 

the usual conceptualization. Three additional concepts were proposed 

as necessary for the arousal of satisfaction. Demonstration of the 

necessity of two of these concepts comprises the empirical portion of 

the dissertation. 

In this chapter, research hypotheses are derived from the model 

and theory that have been discussed. These general hypotheses are 

then interpreted in the context of a specific research methodology in 

Chapter 5. In the interest of clarity, the discussion will continue to 

focus on satisfaction within the consumer domain. 

Main Hypotheses 

Hypothesis 1a : When all necessary conditions are fulfilled 
(i.e., expectations, decision, experience, perception, high 
response involvement and disconf irmation are all present), a 
positive discrepancy (i.e. perception surpasses expectations) 
results in the pattern of emotions that constitutes 
satisfaction . 



167 



168 



Hypothesis 1b : When all necessary conditions are fulfilled, 
a negative discrepancy (i.e. perception falls short of 
expectations) results in the pattern of emotions that 
constitutes dissatisfaction (negative satisfaction) . 

These hypotheses re-state the basic premise of the prevailing 
disconf irmation paradigm in terms that reflect the modified 
conceptualization being presented. In addition to the emotional 
measures, more standard measures of satisfaction (Figure 4) should also 
demonstrate standard results. When one or more of the necessary 
conditions is unfulfilled, however, there will be a divergence among 
the measures. 

As discussed in the previous chapter, the act of making a decision 
serves to solidify expectations and to make the object of the decision 
more personally relevant. Following the logic of hypotheses 1a and 1b, 
a lack of adequate emotional response in conditions in which decision 
was absent comprises support for a critical test of the necessity of 
decision in the satisfaction process. This test is founded on the 
distinction between satisfaction as a pattern of emotions and 
satisfaction as the response to a purported satisfaction scale. 
Chapter 3 described the motivational nature of emotions and argued that 
one source of the motivation necessary for experiencing the 
satisfaction emotion is response involvement. A logic similar to that 
described above suggests that it should be feasible to prevent 
satisfaction responses by creating a situation that entails low 
response involvement. At the present level of knowledge, prediction of 
how much response involvement is sufficient necessarily must be 
tautological (i.e., a sufficient level is associated with an emotional 



169 



response whereas an insufficient level does not evoke emotion). 

However, once again it is emphasized that standard satisfaction scales 

are not sensitive to the lack of emotion. 

Hypothesis 2 : Commonly used measures of satisfaction do not 
discriminate the absence of satisfaction or dissatisfaction in the 
presence of a perception-expectation discrepancy. That is, the 
usual measurement instruments are insensitive to the 
non-fulfillment of such necessary conditions as decision and 
sufficient response involvement. 

Hypothesis 3 : Measures of attitude do not discriminate the 
absence of satisfaction or dissatisfaction as results from 
the omission of a decision or adequate response involvement. 

The hypotheses in this group are predicated on the major argument 

presented in the dissertation: the use of standard satisfaction 

measures can yield responses that are better described as measurement 

artifacts than as the result of the consumption situation. In effect, 

omission of one or more of the necessary conditions allows for a 

detached, cognitive evaluation of the target, but will not arouse the 

emotions that constitute the satisfaction response. Attitude measures 

and the satisfaction scales from prior studies will be responsive only 

to the perceptual process that is driven by performance of the target 

object. These do not indicate that satisfaction responses will result 

(see Hypothesis 4, below). 

Secondary Hypotheses 

Hypothesis 4a : Satisfaction, as defined and identified by 
emotional response, is associated with a greater incidence of 
those behaviors (e.g., positive word-of-mouth, complimenting) 
usually thought to follow from the feeling. 

Hypothesis 4b : Dissatisfaction, as defined and identified by 
emotional response, is associated with a greater incidence of 
those behaviors (e.g., negative word-of-mouth, complaining) 
usually thought to follow from the feeling. 



170 



Hypothesis 4c : Attitude toward the brand is associated with 
a corresponding intent to repurchase. 

As discussed at length in Chapter 2, the original goal of the CS/D 

literature was to understand specific behaviors thought to result from 

satisfaction (Cardozo 1965). Chapter 3 presented a conceptualization 

that accounts for the weak relationships found frequently in empirical 

investigations. That is, although attitude will influence repurchase 

intention, it is not sufficiently motivating to initiate the other 

behaviors of interest. 

Hypothesis 5 : The likelihood of engaging in 

satisfaction-related behaviors increases with the strength of 
the satisfaction emotions, whether positive or negative. 

This hypothesis operationalizes the argument depicted in Figure 6, 

that the different behaviors require different levels of motivation. 

Chapter 2 pointed out that these motivational levels may be dependent 

on the strength of the satisfaction feeling. Previous attempts to 

establish the correlation between level of satisfaction and behavioral 

response have been handicapped by the lack of a sufficiently sensitive 

satisfaction measure (Richins 1983b). The present hypothesis assumes 

an adequately sensitive measure and that perceptions regarding the 

relative cost and benefit and the appropriate mechanism for completing 

each response are constant across subjects (Day 1984). 

Supplemental Hypotheses 

Hypothesis 6 : Undertaking a decision process directly 
influences perception of the outcome of the decision. When a 
decision has been made, variance from expectations is 
exaggerated. 

This hypothesis summarizes the basic contention of the dissonance 

literature that was adopted by the early consumer satisfaction 



171 

researchers. That is, Anderson, Cardozo and others were interested in 

the effects that purchase and consumption would have on one's ability 

to veridically perceive product performance. In this case, it is 

expected that any misperception that results would be consistent with 

assimilation-contrast theory (Sherif and Hovland 1961). In Chapter 3, 

it was argued that the personal risk aroused by decision is the most 

important cause of any shift in perception. 

Hypothesis 7 : Perception of a product (performance) has a 
direct influence on attitude toward that product. 

The CS/D literature has concentrated rather strongly on the 

relationships among perception, disconf irmation and satisfaction. 

Attitude, on the other hand, is viewed as affected by satisfaction 

alone (Figure 3). The current conceptualization reflects a different 

perspective; attitude can be influenced by either perception or 

feelings of satisfaction or both (Figure 8). This reflects Oliver's 

findings that suggest postusage satisfaction is influenced by the 

disconf irmation judgment which was based on perception gained from 

personal experience (1980a). Thus, the present model is more 

responsive to criticisms that objectively "good" or "poor" performance 

should not be totally counter-balanced by prior expectations. 

Hypothesis 8 : As response involvement increases, strength 
of emotional response increases. 

Again, there is a ceteris paribus condition implicit in the 

hypothesis. Hypotheses 5 and 8 provide a test of the mini-theory of 

"Conservation of Motivation" discussed earlier. The role of motivation 

in emotion research was elaborated in Chapter 3 and shown to be a 

critical distinguishing feature between satisfaction and attitude. The 



172 
present conceptualization explicitly links origins (response 
involvement) and outcomes (behavior) of motivation. 

Summary 

The dissertation presents an attempt to integrate a number of 
theories relevant to satisfaction into a coherent conceptualization. 
The theoretical network that is generated can not be evaluated in toto, 
but must be tested in parts. The hypotheses presented in this chapter 
constitute the initial phase of such testing. 

The main hypotheses are identified as such because the primary 
orientation of the dissertation is the definition of the construct and 
the presentation of a prototypical satisfaction situation. Thus, 
evaluation of the necessity of newly introduced constructs (in the 
present study, decision and response involvement) in the context of the 
definition adopted is imperative. 

The secondary hypotheses concern other aspects of the 
conceptualization which can be examined as part of the research that is 
envisaged, but which were not central to the initial design. As such, 
evidence gathered pertaining to these hypotheses should be interpreted 
as suggestive, in lieu of that which could be collected in studies 
specifically designed to investigate these issues. 

Finally, the supplemental hypotheses were drawn from arguments 
that have been raised in the CS/D and related literatures. The 
hypotheses, as stated, are consistent with both the present 
conceptualization and the prevailing wisdom of the CS/D literature. 
Contrary findings, however, would not be interpretable as damaging to 
the central model. As with the secondary hypotheses, the supplemental 



173 
hypotheses represent interesting, albeit tangentially important, 
portions of the overall theory. 

There are quite a number of issues raised in the literature review 
and conceptualization chapters that are not addressed by the empirical 
portion of the dissertation and are therefore not represented among the 
hypotheses. The investigation of what is essentially a network of 
theoretical constructs is necessarily an iterative process. A refining 
of theories and measures is implicit in any research stream, of which a 
study such as this is only the beginning. A plan for examining some of 
these issues is presented in more detail in Chapter 7. 



CHAPTER 5 

EXPERIMENTAL METHOD 

Introduction 

The concept of satisfaction has been defined as a general pattern 
of emotions that may arise in any of several domains. A general 
framework that relates the construct to its antecedents, outcomes and 
associated concepts has been presented in some detail. In the present 
chapter, the definition and conceptualization are operationalized and 
evaluated in light of specific experimental hypotheses derived from 
more general hypotheses presented in Chapter 4. 

The present chapter begins with a discussion of the development of 
the measures that will be used in the main study of the dissertation. 
The emotional measures that are derived constitute an operational 
definition of the construct that was tentatively defined in Chapter 3 
(see also Figure 7b). The behavioral intention scales, as derived, are 
also considered to be measures of satisfaction, albeit indirect since 
they actually assess theoretical outcomes (Figure 9). Finally, the 
complete experimental methodology is presented with special emphasis on 
the manipulation of the proposed antecedent conditions. Operational 
versions of the hypotheses presented in Chapter 4 conclude the present 
chapter . 

Pilot Study--Measure Development 

Westbrook's efforts to describe the patterns of emotions that 
arise during ownership (1983; Westbrook and Oliver 1984) have been 

174 



175 
discussed previously and led to the proposed identification of the 
endpoints of the satisfaction continuum in terms of specific emotions. 
These studies were not intended for, nor were the measures taken 
adequate for, the derivation of such definitions. Therefore, a pilot 
studv was conducted to validate or revise the definitions that have 
been proposed. 

The studies reported by Westbrook provide no criteria by which to 
identify the emotional profiles that best describe satisfaction. The 
pilot study corrects for this in two ways. (1) The critical incident 
technique that has been introduced to the CS/D literature (Maddox 
1981; Swan and Combs 1976) is used to concentrate responses at the ends 
(negative and positive) of the satisfaction continuum. (2) Measures of 
intent to engage in specific behaviors thought to result from 
satisfaction (as discussed in Chapter 2) are collected in addition to 
the more typical satisfaction scales used by Westbrook. 

A concise framework for the construction of psychometric scales 
has been presented in the marketing literature and can be adapted for 
use in the present study (Lundstrom and Lament 1976). The principles 
and philosophy implicit in the scale derivation are consistent with 
those of the author and those presented in other recent discussions of 
measure development (Churchill 1979; Nunnally 1978). The Lundstrom and 
Lamont framework presents five major considerations for the 
construction of a reliable and valid scale: (I) definition of the 
construct and selection of an item pool; (2) selection of a scaling 
procedure; (3) selection of scale items; (4) evaluation of the 



176 
reliability of the scale; and (5) evaluation of the validity of the 
scale. 

The pilot study examines two methods of measuring satisfaction, 
defined previously as "an emotional response . . . conceptually 
distinct from . . . behavioral responses" (Day 1933). One set of 
measures taps directly the emotional nature of the construct. The 
second set of measures discussed examines intent to behave in a 
particular manner. Although this latter set of scales, by definition, 
can measure the construct only indirectly, the model presented 
previouslv indicated the importance of behavioral response to the 
overall conceptualization. 
Obj ectives 

The principal goals of this study were to identify those emotional 
measures and those measures of behavioral intent that best discriminate 
between subjects responding to satisfying and dissatisfying occasions, 
respectively. Satisfaction scales and criteria for evaluation can then 
be constructed from these measures. 
M ethod 

Ninety-four students in an introductory Marketing class served as 
volunteer subjects for extra class credit. The study was conducted in 
various sized groups in a classroom setting and required about thirty 
minutes per session. 

Subjects were given a blank sheet of paper on which to first 
identify a recent consumer experience that was particularly 
dissatisfying and then, after all had completed this, identify a recent 
consumer experience that was particularly satisfying. (This order was 



177 
chosen because it was felt that a satisfying occasion would be more 
readily recalled if "primed" by an already salient dissatisfying 
occasion.) After sufficient time for all subjects to identify an 
occasion of each type, they were instructed to turn to and complete 
the first page of the questionnaire (Appendix A). On this page, they 
were to provide more details concerning either the satisfying (N=46) or 
dissatisfying (N=48) occasion. 1 

In order to avoid hastily completed and unnecessarily brief 
descriptions of the specific occasions, subjects were instructed not to 
continue through the questionnaire until sufficient time had elapsed 
for a complete narrative (a pre-test indicated that 10 minutes was an 
adequate period). The remainder of the questionnaire was self-paced 
and consisted of emotion scales, satisfaction measures and behavioral 
scales (the entire questionnaire is contained in Appendix A). The 
narratives were required in order that subjects would revisit the 
emotional experience (Robinson 1976, 1980; Weiner, Russell and Lerman 
1979). 
Measurement scales 

Westbrook employed an abbreviated version of Izard's Differential 
Emotions Scale, "owing to the length of the original scale (69 items)" 
(1983, p. 5). Yet, he chose not to use the abbreviated version that 
had already been supplied by Izard (1977). The shortened forms used by 
each of these authors have been discussed and are presented in Table 3. 



1 In the course of analysis, it was determined that four subjects 
in the satisfying condition actually described, and completed the 
remainder of the study responding to, the dissatisfying occasion. 
This problem and isolated incidences of item non-response created some 
variations in sample size, as reported below. 



Due to a lack of rational criteria by which to select between these 
and, in the face of Izard's estimate of 5 minutes as the time 
requirement for a 67-item edition of the scale (1972, p. 86), the full 
version was administered. In addition, adjectives were extracted from 
satisfaction scales to round out a 75-item list (added were: 
Dissatisfied, Satisfied, Pleased, Contented, Thrilled, Depressed, Wise, 
Foolish) . A single randomized order was reproduced on all 
questionnaires, but individual subjects were assigned separate starting 
points and instructed to complete the entire list. 

Four satisfaction scales were selected from the assortment 
presented in Figure 4. The scales were chosen on the basis of 
reliability in prior testing (Oliver and Westbrook 1982; Westbrook and 
Oliver 1981) and production feasibility (i.e., verbal rather than 
graphic scales) . 

The questionnaire concluded with a series of Likert type scales. A 
form of Oliver's (1980a) six-item satisfaction scale began the list and 
was followed by a series of behavioral intent scales (phrased as 
behavioral likelihood "I would ..." "Strongly Agree--Strongly 
Disagree"). This latter set was derived from a variety of sources 
including satisfaction surveys, guides for dissatisfied consumers and 
personal experience. The behavioral intent measures were intended to 
be exploratory in nature. 
Analysis 

The first of Lundstrom and Lamont ' s (1976) considerations has been 
satisfied; definition of the construct and identification of the item 
pool(s). The selection of Izard's DES as the basis for emotional 



179 



measures identified the scaling procedure as well as the item pool. 
Thus, a Likert-like scale will be used to measure emotions. For 
reasons similar to those elaborated by Lundstrom and Lamont 
(feasibility, simplicity), a Likert (1932) scaling procedure was also 
selected for the behavioral intent items. Comparisons of Likert and 
other procedures have generally concluded that the scales which result 
from the former are at least as reliable (e.g., Thurstone's technique, 
Edwards and Kenney 1946). 

The major task of the Pilot Study involved the selection of 
specific items as measures of the satisfaction construct. Several 
analyses of the emotion scales were conducted in order to identify 
those emotions which best characterize each end of the satisfaction 
continuum. The 75 items that constituted question 4 were subdivided 
four ways: items used by Westbrook (1983); items used by Izard (1977) 
in a shortened scale; those items that exhibited the highest factor 
scores in Izard's factor analysis; and the full scale. 

In general, the analyses conducted were of the following types: 

Factor Analysis of Emotions — A varimax principal components 
factor analysis was conducted for each of the four sets of 
emotion measures. These analyses were performed to ascertain 
whether the adjectives said to describe separate emotions 
would load appropriately. Also, given the restricted 
emotional patterns expected (i.e., only satisfaction and 
dissatisfaction), this analysis was expected to suggest which 
emotions might combine in each pattern. 

Discriminant, Cluster and Regression Analysis of Emotions- 
-Scales were created by summing the items loading on each 
factor resulting from the prior analyses (Table 8). These 
intermediate scales were then employed in efforts to segment 
subjects. Discriminant analysis using assigned condition 
(satisfying versus dissatisfying occasion described), K-means 
cluster analysis looking for "good" two-cluster solutions and 
regression of more standard satisfaction measures on the 
emotional scales were expected to result in identification of 



Table 8 
Results of Factor Analysis of Emotion Items 



I. Using items from Izard (1977) 

Factors and loadings 
Emotion Item 1 2 3 4 5 6_ 

Joy 



Distress 
Disgust 

Anger 
Contempt 



Interest Attentive .8747 

.7696 
.7486 
Fear Scared .9404 

,8326 
,9225 
Surprise Surprised .8191 

.8274 
.7724 
Shyness Sheepish .5707 



Guilt 



Delighted 


-.7144 


Happy 


-.7839 


Joyful 


-.6470 


Downhearted 


.7205 


Sad 


.5889 


Discouraged 


.6968 


Feeling of 




Distaste 


.8093 


Disgusted 


.9016 


Feeling of 




Revulsion 


.8139 


Enraged 


.9077 


Angry 


.9221 


Mad 


.9181 


Contemptuous 


.6923 


Scornful 


.8528 


Disdainful 


.7441 


Attentive 




Concentrating 




Alert 




Scared 




Fearful 




Afraid 




Surprised 




Amazed 




Astonished 




Sheepish 




Bashful 




Shy 




Repentant 


.4651 


Guilty 




Blameworthy 





.8428 

.7701 



.7851 
.8237 



181 
Table 8--continued 



Results from Westbrook's Items (29 items) 

Factor 1 Distress, Disgust, Anger, 

Contempt 

Factor 2 Interest, Joy 



Factor 3 Surprise 

Factor 4 Guilt, Shyness 

Factor 5 Fear 

Factor 6 Shame 



Results from items selected from Izard's Factor Analysis (49 items) 

Factor 1 Joy, Distress, Disgust, Anger 

Factor 2 Interest 

Factor 3 Fear 

Factor 4 Surprise 

Factor 5 Shyness 

Factor 6 (?') a 

Factor 7 Guilt 



Results from all items (75 items) 



Factor 


1b 


Distress, 


Factor 


2^ 


Joy 


Factor 


3 


Fear 


Factor 


4 


Surprise 


Factor 


5 


Interest 


Factor 


6 


Shyness 


Factor 


7 


Guilt 


Factor 


8 


(?) 


Factor 


9 


(?) 


Factor 


10 


(?) 



Disgust, Anger, Contempt 



Note: a (?) indicates that no clear characterization could be made 
b Several of the added items (satisfied, dissatisfied, etc.) 
loaded on these. 



182 



the emotions most representative of satisfaction and 
dissatisfaction. 

Factor Analysis and Other Analyses of Behavioral Intent Measures 
— Rather than attempt to measure and scale actual behaviors in the 
Pilot Study, Likert scales of potential satisfaction related 
behaviors were used as a surrogate. As with the emotional scales a 
varimax principal component factor analysis was used to derive 
intermediate scales that were used in further analyses (Table 9) . 
Discriminant analysis using assigned condition and mean comparisons 
across groups defined by emotional profiles were conducted to 
identify measures of behaviors most typical of satisfied and 
dissatisfied respondents. 

Results 

Given the exploratory nature of the study, the number and selection 
of subjects, the lack of blind judging and the intentional selection of 
extreme results, it would be misleading to conduct or report 
significance tests for this study. Where statistics are shown (e.g., 
Cronbach's alpha), these indicate criteria used in selecting items for 
further analysis. As a pilot study, the primary result is intended to 
be the actual measurement scales rather than statistics relating to 
their reliability and/or validity. Such estimates would have to be 
based on a sample separate from that used in derivation of the scales 
(Cureton 1950) . 
Emotion scales 

In general, the factor analyses of emotional responses provided 
encouraging results. By and large, the adjectives that were predicted 
to describe a particular emotion did load on the same factor. 
Furthermore, the emotions that had been suggested in Chapter 3 as 
components of satisfaction and dissatisfaction also tended to load 
together (Table 8). As indicated above, the mean score on each scale 
listed in Table 8 was used in further analyses. 



183 



Table 9 
Behavioral Intent Scales from Factor Analysis 



Item # from Factors and Factor loadings 

Appendix A Description 1 2 3 4 5 6 



1 Satisfied with 
decision to purchase/ 

use* .8723 

2 If do over, would 

feel differently 3 -.6750 

3 Choice was wise a .8467 

4 Feel bad about 

decision 3 -.7634 

5 Did right thing a .7607 

6 Not happy that I 

did what I did 3 -. 6378 

7 Would buy other 
products of same 

brand .5547 

14 Tell a friend good 

things .7530 

20 Recommend product to 

a friend .7890 

21 Recommend friend 

avoid product -.4811 .4720 

22 Publicly endorse 

product .5987 

12 Complain to higher 

authority .8511 

15 Tell a friend bad 

things .7753 

16 Write manufacturer 

and complain .8353 

23 Publicly attack 

product .5387 



184 



Table 9 — continued 



Item // from 
loadings 
Appendix A 



Description 



Factors and Factor 

3 4 5 6 



18 Discard and replace 
with different brand 

19 Discard and replace 
with different product 

8 Buy other products 
from same seller 

13 Compliment higher 

authority 
17 Write manufacturer 

and compliment 

10 Buy more of product 
as gift 

11 Buy similar product as 
gift 

9 Buy other products 
of same type 

24 Fight to remove 

product from market 



Number of Items 
Coefficient a 

(based on Pilot) 



11 4 
.95 .83 



7860 








7823 


.5869 

.7417 
.81 13 


.8811 
.9058 


7637 






- 


5973 


3 


3 


2 


2 


.85 


.72 


.91 


.47 



Note: a From Oliver (1980a) 



135 
The emotional scales were used to predict whether the subject had 
been in the satisfied or dissatisfied condition of the experiment 
(discriminant analysis using PROC GLM from the Statistical Analysis 
System) and to predict the subject's responses on both the more typical 
satisfaction scales and the behavioral intent summed scales (regression 
and correlation in SAS) . Two different cluster analysis programs (SAS 
and BMDP — K-means) were used to describe subsets of the sample. 

No single analysis led to firm conclusions. However, the bulk of 
the evidence suggested the following conclusions: 

1. Izard's abbreviated DES (1977) was more reliable (Table 
10) and yielded more consistent results than either 
Westbrook's version or scales developed within the 
present study. 

2. The emotions that were most characteristic of persons in 
the "satisfied" condition and were most associated with 
highly positive satisfaction and behavioral intent 
responses were Interest, Joy and Surprise (Table 11). 

3. The emotions that were most characteristic of persons in 
the "dissatisfied" condition and were most associated 
with highly negative satisfaction and behavioral intent 
responses were Anger, Disgust and Surprise (Table 11). 

These emotions, then, comprise working definitions of either end 

of the satisfaction continuum. The satisfaction end of the scale, at 

least, is consistent with the conclusion drawn by Westbrook, "joy and 

interest, once elicited in a consumption experience, are necessarily 

involved in behavior relevant to the stimulus . . . the affect of 

surprise also is likely to be involved" (1987, p. 259). 2 Operational 

definitions were derived based on mean responses by subjects in each 

condition. For the main study of the dissertation, a satisfaction 



2 Westbrook distinguishes the effect of surprise as an 
"amplifier" of other emotions. This suggests that some different 
algebraic representation of surprise versus other emotions--e . g. , 
multiplicative rather than additive, respectively, may be in order, 



186 



Table 10 
Reliability Coefficients for Scales Used in Pilot Study 



Interest 

Joy 

Surprise 

Sadness 

Anger 

Disgust 

Scorn 

Fear 

Shame 

Guilt 



Number C 
Of Items a 


oef f ici 
a 


3 


.80 


3 


.95 


3 


.80 


3 


.77 


3 


.95 


3 


.89 


3 


.83 


3 


.90 


3 


.60 


3 


.59 



Note: a Based on Differential Emotions Scale (Izard 1977) 



137 



Table 11 
Adjectives Describing Ends of Satisfaction Continuum 



Satisfaction: 
Emotion Adjectives 

Interest Attentive, Concentrating, Alert 

Joy Delighted, Happy, Joyful 

Surprise Surprised, Amazed, Astonished 

Dissatisfaction: 

Emotion Adjectives 

Anger Enraged, Angry, Mad 

Disgust Feeling of Distaste, Disgusted, Feeling of 

Revulsion 

Surprise Surprised, Amazed, Astonished 



pattern of emotions will consist of an average response equal to or 
greater than 3 (a response of "Moderately") on the items comprising the 
characteristic emotions. Similarly, dissatisfaction (negative 
satisfaction) is defined as an average response equal to or greater 
than 3 on items specific to that pattern of emotions (Table 11). 

When these definitions were compared with the responses of the 94 
subjects, 36 met neither cut-off, 7 met both cut-offs, 50 were 
correctly classified into original condition and 1 subject from the 
dissatisfied condition was identified as satisfied. All of the 
"misclassif ied" subjects were from the dissatisfied condition of the 
study. Further examination showed that there was a tendency to report 
"dissatisfying" occasions that were resolved in the subject's favor 
and resulted in a net satisfaction. This explanation is corroborated 
by the responses on classical satisfaction scales. 

Those subjects who met neither cut-off (14 from the satisfied 
condition and 22 from dissatisfied) represent a different problem for 
the measurement technique. Either the definitions that have been 
constructed are not sufficiently sensitive, or these subjects were not 
satisfied (or dissatisfied) as defined by the conceptualization offered 
by the dissertation. Although not conclusive, some insight can be 
gained by comparison against the antecedents and resulting behavioral 
intentions . 

Of the 36 respondents who met neither set of criteria, only three 
did not describe a situation which encompassed all six antecedents; 
nine did include some mention of the antecedents and 24 were counted as 



most likely satisfying the antecedents. 3 These results indicate that 
caution is required in interpretation of the sufficiency of the 
antecedent conditions. That is, although an experience that fulfills 
the conditions should position a person along the satisfaction 
continuum, it may not be sufficient to arouse strong emotions. Thus, 
the 33 subjects who met the antecedent conditions but not the emotional 
criteria would be positioned in more central portions of the continuum. 
Description of this region is beyond the scope of the present study and 
not appropriate for the data collected. Future studies based on the 
conceptualization derived in the dissertation can make specific 
definitions and predictions regarding the middle range of the scale and 
at that time establish operational definitions based on emotional 
patterns . 
Behavioral intent measures 

One advantage of the present conceptualization is the integration 
of specific results (behavioral, motivational) into the overall 
satisfaction theory. This enables the use of behavioral intent 
measures as well as satisfaction scales to compare subjects who have 
been classified by emotional response. 

Table 12 summarizes the mean responses to various dependent 
measures by experimental condition or subsequent emotional 
classification of subjects. There are a couple of points notable from 
the table: (1) The difference between survey groups (conditions) 



3 The narratives provided by the subjects were scored by the 
author and required judicious interpretation, as with the early pilot 
findings reported. This procedure is adequate for the exploratory 
nature of these particular studies. 



190 



Table 12 
Mean Responses by Condition and Classification 



Condition Classification 
Satisfied Dissatisfied Satisfied Dissatisfied 

Measure (n=42) (n=50) (n=29) (n=29) 

Dependent Measure 15 

Scale U 4.35 2.53a 4.39 2.37a 

Scale 2 3.70 2.64a 3.73 2.52a 

Scale 3 4.27 2.96a 4.34 2.89a 

Scale 4 3.31 2.41a 3.46 2.19a 

Scale 5 2.96 2.30b 3.19 2.11* 

Scale 6 4.12 3.70^ 4.14 3.52- 
Satisf ied- 

Dissatisf ieds 6.79 1.77a 6.62 1.69a 

Odds 9.29 2.48a 9.24 2.00a 
Percent 

Satisfied 93.10 20.96a 93.45 17.59a 
Deiighted- 

Terrible 6.40 2.43a 6.59 2.31a 



Note: Significance of results is indicated by a = p< . 0001 ; b = p< -01; 

c = p<.05; d = p< . 10. 
- ■ Some scales have been reversed so that higher numbers are more 

positive . 
f "Scales" are mean responses to behavioral intent scales derived 

from Table 9. 
§ Satisfaction measures are from question 5 of Appendix A, parts a 

through d. 



191 
resulted in significantly different responses across all measures. (2) 
As shown in Appendix A, the Likert type scales used to assess 
behavioral intention were centered at Neither Agree nor Disagree with a 
statement indicating what the subject might "normally do" under the 
circumstances described. The responses to the items comprising summary 
scales four, five, and six in Table 12 lie fairly close to this 
response, that had been assigned a value of three, and therefore are 
not likely to discriminate well between satisfied and dissatisfied (or 
indifferent) individuals. (3) The emotional classification procedure 
generated slightly more extreme means than those collected from the 
entire sample. 

While the summed scales are adequate to represent differential 
responses on the items, it is also of interest to determine which 
individual items may be useful in the main study. Table 13 summarizes 
the mean values for each of the dependent measures, by condition and by 
emotional classification. Those items that exhibit high absolute 
differences are candidates for inclusion as behavioral intent or 
behavioral measures in the main study of the dissertation. 4 

Overall, behavioral intent, emotion and more standard measures or 
satisfaction have tended to converge on the distinction between 
groups. At this juncture, however, it would be inappropriate to 
attempt to compare the measures competitively. There is no reason to 



'- There is a level of abstraction problem raised by blind 
acceptance of these results. The situations described by the subjects 
in the pilot study involved a variety of products and services which 
may have aroused emotions focussed at any of the levels suggested by 
Table 2. For a given target of the emotion, behaviors specific to 
that target should be most affected, ceteris paribus. 



192 



Table 13 
Mean Responses to Likert Items by Condition and Classification 



Condition Classification 

Satisfied Dissatisfied Satisfied Dissatisfied 
Measure*.* (n=42) (n=50) (n=28) (n=29) 



2.08a 4.72 1.89a 

2.68a 4.21 2.54a 

2.32a 4.55 2.07a 

3.10a 4.69 3.07a 

2.86a 4.52 2.61* 

3.26a 4.7 2 3.14* 

2.76a 4.21 2.54a 

2.92a 4.00 2.50a 

3.60 3.86 3.36 

2.16^ 3.24 1.96'° 

2.44 3.14 2.25= 

2.46a 3.93 2.21a 

2.14a 3.48 2.07^ 

2.14a 4.24 2.04a 

1.96a 3.31 1.86a 



1 


Decision 








satisfaction 


4 


67 


2 


Feel Differently 


4 


26 


3 


Wise choice 


4 


55 


4 


Feel bad 


4 


69 


5 


Did right thing 


4 


50 


6 


Not happy 


4 


71 


7 


Buy brand 


4 


10 


8 


Buy from seller 


3 


93 


9 


Buy product type 


3 


76 


10 


More gift 


3 


00 


11 


Similar gift 


2 


93 


12 


Complain higher 


3 


93 


13 


Compliment higher 


3 


31 


14 


Friend/good 


4 


21 


15 


Friend/bad 


3 


07 



193 
Table 13 — continued 

Condition Classification 

Satisfied Dissatisfied Satisfied Dissatisfied 

Measure^ , f (n=42) (n=50) (n=28) (n=29) 

16 Complain 

manufacturer 3.71 3.00^ 3.52 2.96 

17 Compliment 

manufacturer 2.69 2.18= 2.90 2.00^ 

18 Replace brand 4.24 2.90 a 4.31 2.82* 

19 Replace product 4.31 3.02 a 4.38 2.96* 

20 Recommend buy 4.29 2.20 a 4.38 2.11* 

21 Recommend avoid 3.98 2.32 a 4.00 2.25 a 

22 Endorse 3.55 1 . 94 a 3.66 1.71* 

23 Attack 4.10 3 . 1 4 a 4.17 3.04^ 

24 Ban 4.48 3.80 b 4.41 3.68^ 



Note: Significance of results is indicated by: a = p<.0001; 

b = p< .01 ; c - p<-05; d = p< . 10. 
e Some scales have been reversed so that higher numbers are 

more positive. 
f Measure numbers correspond to item numbers in Table 9. 



194 
expect divergence in any particular direction, nor was the study- 
designed to provide such tests (a replication should indicate 
convergence among the measures). Moreover, the scales were developed 
iteratively using one another as criteria, so no true indicants of 
reliability are available. Likert type scales were derived with 
procedures similar to those usually prescribed. Of special note is 
that the situations were specifically chosen to create extreme 
responses on the various measures. Thus, the design of the study 
selected the extreme groups to use in measure selection, as is 
generally recommended (Edwards and Kenney 1946; Likert 1932; Shaw and 
Wright 1967) . 

The behavioral intent measures may be useful in developing 
measures that are based on actual behavior. Behavioral measurement of 
attitude has been attempted, but involves considerable difficulty and 
cost to execute and, due to the other forces acting on behavior, may be 
of questionable reliability (Cook and Selltiz 1964; Fishbein 1966). it 
has been suggested that weighting behaviors may make for more reliable 
measures (Rosander 1937), but such weighting schemes tend to be 
unstable (Triandis and Triandis 1965). The best option would seem to 
be to collect both intent and actual behavior measures and look for 
convergence . 5 



5 An assumption implicit in the discussion of behavioral measures 
is that the experimental design will control such factors as cost and 
effort required for each behavior and, to an extent, expected result. 
That is, it is expected that perceived costs and benefits of various 
responses would be equivalent across subjects; only motivation would 
vary. This mitigates the arguments presented in the CS/D literature 
against complaints as a measure of satisfaction. 



195 
Summary of the pilot study 

The study used the critical incident technique that has been used 
successfully in the job satisfaction literature to revive the memory of 
particularly salient emotional events. Emotional definitions of the 
positive and negative endpoints of the satisfaction continuum were 
derived and shown to be feasible to ascertain with a simple 
questionnaire. The study also indicated behavioral intent measures 
that discriminated well between groups of subjects. The results from 
this study are used in deriving both measures and hypotheses for the 
major empirical portion of the dissertation. 

Research Conducted 
Purpose 

Having defined the satisfaction and dissatisfaction constructs as 
specific patterns of emotions, it is necessary to investigate the 
conditions that are hypothesized causes of the feelings. The study 
evaluated two of the three concepts that have been added to the 
conceptualization of satisfaction and whether each is necessary to the 
arousal of emotions. The measures derived in the Pilot Study just 
reported are used in conjunction with measures of behavioral intent and 
more traditional satisfaction measures to examine the appropriateness 
of the derived conceptualization. 
Design 

For the purpose of providing as clean a test of the hypotheses as 
possible, satisfaction with a single purchase/use occasion provided the 
focus of the study (Table 1). Also, the critical tests will focus on 
immediate response to the situation, before motivation begins to decay. 



196 



Six antecedent conditions have been proposed as necessary and 
collectively sufficient to arouse satisfaction emotions. The presence 
of two of the three introduced by the current conceptualization-- 
decision and high response involvement — are manipulated factorially. 
The discrepancy between expectations and perception (perceived 
performance) is manipulated to either a strongly positive level (i.e., 
low expectations and good performance) or a strongly negative level 
(i.e., high expectations and poor performance). Thus, the overall 
design is a 2x2x2 factorial (the disconf irmation judgment is expected 
to occur when all of the other antecedent conditions are fulfilled) . 
Target Product 

The design of the experiment made selection of the product to be 
used somewhat difficult. Several criteria were identified that needed 
to be met in order to conduct an effective and credible experiment. 

1 ) The experimenter must be able to manipulate 
product performance in a simple manner that will 
yield a consistent and unambiguous perception. 
Yet, the manipulation of performance must be 
sufficiently subtle to avoid raising response 
involvement inadvertently. 

2) The subject must be able to assess performance 
after a single trial which can be performed within 
an experimental setting. 

3) Response involvement is dependent on enduring 
involvement which must therefore be fairly low in order 
to achieve the low involvement condition. On the other 
hand, the situation must be able to raise involvement 
credibly to a level sufficient to drive the satisfaction 
process . 

4) The product must be useful to the subjects in 
their ordinary lives so that pre- and post-trial 
choice measures are credible. 

5) Similarly, prior consumption within the product 
class will help to establish more widely held norms 



;97 



for performance. Specific expectations, however, must 
still be subject to manipulation. 

6) The overall procedure must be credible to the 
subject. That is, a condition requiring no 
decision and low response involvement would be 
suspect without a viable "cover story". 

As discussed, ground coffee (for automatic drip cof f eemakers) was 

chosen from a candidate list based in part on the products used in 

prior CS/D research (Table 7) and those discussed by respondents to 

pilot studies. Coffee was used in two early investigations (Cohen and 

Goldberg 1970; Olson and Dover 1976) and fits the criteria quite well. 

1) Coffee has a limited number of salient 
attributes which suggests that performance 
perception can be manipulated via only a limited 
number of flavor dimensions such as strength of 
flavor and bitterness (Eastlack 1964; Olson and 
Dover 1976). The flavor quality of coffee can be 
lowered to a poor, but acceptable, level by a 
variety of techniques (e.g., use of a poor tasting 
additive) . 

2) In general, food products can be consumed and 
evaluated rather quickly. 

3) Prior research suggests that a study which is 
focused on coffee selection can be relatively 
involving. On the other hand, coffee is not so 
intrinsically involving that a diversion would be 
unsuccessful . 

4) The subjects, whether students or a more general 
sample, can be screened to identify coffee 
drinkers . 

5) Coffee drinkers would be expected to have some 
expectations of normal flavor. Such subjects would 
be expected to have ample prior experience so that 
product class norms should dominate any comparison 
level judgments, rather than a specific alternative 
choice (Woodruff, Cadotte and Jenkins 1983a). 
Specific expectations regarding the experimental 
sample can be manipulated by use of unfamiliar 
brands or brands not yet available. 



198 



6) The study conducted by Cohen and Goldberg 
included many procedures that are adapted to the 
present study (1970). Specifically, they varied 
decision and non-decision, involvement (although 
they were concerned with ego involvement which the 
present conceptualization identifies more with the 
decision manipulation) and a taste manipulation. 

Rather than instant coffee as Cohen and Goldberg used, the study 
undertaken used coffee ground for automatic drip makers. This provided 
the ability to use a coffeemaker as a decoy product. The presence of a 
decoy enables the involvement manipulation to be presented to the 
subjects as a focus of attention or interest. 
Manipulations 

The basic manipulations are described in Table 14. 
Operationalization of these within the design necessitated slightly 
different procedures within each cell; therefore, there were eight 
different protocols (Appendix B) . 

Discrepancy . Central to most CS/D investigations is the 
discrepancy between expectations and perception. As mentioned 
previously, a subject participated in either a positive or a negative 
discrepancy condition (i.e., the present study does not examine the 
result of expectations being confirmed). Each discrepancy condition 
was defined by separate manipulations of expectations and product 
performance. 

The positive discrepancy condition required relatively low 
performance expectations be generated within the subjects. Thus, the 
information given about the brand had to be somewhat negative. 
Conversely, in order to maintain credibility regarding the purpose of 
the study and the product's existence as a competitor in the market, an 



199 



Table 14 
Manipulation Summary 



Discrepancy 3 - Positive: Low expectations were generated by 

advertising copy and subject instructions that 
focused on the inexpensive nature of the new blend. 
High performance perception was to be the result of 
using a premium grade--but not f lavored--ground 
roast coffee. 

Negative: High expectations were generated by 

advertising copy and subject instructions that 
focused on the premium quality of the new blend. 
Low performance perception was to be the result of 
using generic coffee blended with ground roast 
peanuts (80;20, respectively). 

Decision - Present: Subjects were asked to select the coffee that 

would be prepared and sampled. Choice was to be 
made between the target brand and an unspecified 
brand of coffee of a quality appropriate to the 
experimental condition. 

Absent: Although the possibility of sampling either of two 
brands was presented, subjects were assigned to the 
target brand by the experimenter. 

Response Involvement - High: Subjects were asked to participate in a 
test of a new coffee to be introduced soon in the 
local market. Each subject was told he/she would 
receive a "week's supply" of the evaluated brand. 

Low: Subjects were asked to participate in a test 
of a new coffeemaker to be introduced soon in the 
local market. Each subject was told he/she would 
receive a chance to win a coffeemaker in a lottery 
among study participants. 



Note: a Target product was a new ground roast coffee from an 
unnamed processor. 



200 

overall positive message had to be communicated. Therefore, the 

manipulation that was adopted presented the brand as very inexpensive 

but of "adequate" quality. 

This information was communicated to the subjects through multiple 

methods. At some point in the study, each subject in a positive 

discrepancy condition was told, 

"The coffee will be of adequate quality, but is primarily 
intended to be a cheap blend to help consumers cope with high 
prices brought about by the recent shortage. We are helping 
them to identify the minimal acceptable quality for the home 
market" (see Appendix B) . 

In addition, subjects were asked to evaluate "mock-ups" of 
advertising proposed for the new brand (Figure 10). The advertisements 
were prepared by a commercial artist and were designed such that only 
the desired manipulations varied (i.e., she prepared a base 
advertisement on which copy blocks, et cetera could be substituted). 6 
Subjects actually evaluated full-color photo-reproductions of hand- 
drawn originals. Each subject evaluated two coffee advertisements as 
part of a booklet of four (also containing two coffeemaker 
advertisements, Figure 10a, 10b). 

In one advertisement the headline read, "And Now ... An 

Affordable Coffee From " (Figure 10c). The body copy read, 

"A medium quality light coffee made from less 
costly coffee beans. Special roasting, grinding 
and blending of these beans keep processing costs 
at their lowest possible level. And there's less 
bitterness than you get with other economy brands. 

Look for coffee to be less costly than even 

your store's brand." 



6 The advertisements were drawn by Maria T. Futules of Akron, 
Ohio. 



201 
The only can pictured on which labelling was readable identified the 
coffee as "Special Blend". 

The other advertisement featured differences only in the body copy 
and an accompanying cents-off coupon (Figure 10d). For the low 
expectations condition, the redemption value of the coupon was ten 
cents and the copy read, 

"originator of the automatic drip coffeemaker is proud to 
present an economical coffee, up to $1.00 less expensive per 
pound than even store brands. You still get acceptable 

quality in every cup. coffee provides delicate flavor 

and a light, clear brew. There's even less bitterness than 
other economy brands." 

As before, the graphic portion of the advertisement did not vary by 

condition. 

Having used these devices to create low expectations, it was 
necessary to find a product which would provide good performance and, 
therefore, a positive disconf irmation. The brand selected was "Uncle 
Charlie's Mountain Rich Coffee (see description in Figure 10e). 7 A 
preliminary taste test was accomplished by preparing and serving some 
of the coffee to faculty, staff and students in the College of Business 
Administration at the University of Akron. All agreed that the coffee 
was of good quality, at least equal to that normally prepared and sold 
in the college (although the normal brand varies). 

The negative discrepancy conditions required that high 
expectations be met with poor product performance. As before, 
expectations were generated through multiple communication modalities. 
Instructions read to the subjects were as follows, 



Uncle Charlie's Products Inc. Grand Rapids, MI 49508. 



202 



"The coffee will be very high in quality but only 
slightly more expensive than present top-of-the- 
line coffees. We are helping the company identify 
the quality most preferred by consumers" (see 
Appendix B) . 

Advertising copy was also written to generate high performance 
expectations. Replacing the copy in Figure 10c was, 

"A delicious full bodied coffee made from the 
finest beans available. Special roasting, grinding 
and blending of these beans bring out the finest 
aroma and taste of the coffee. And there's none of 
the bitterness of most brewed coffee. Look for 

coffee to be priced with your store's medium 
brand coffees." 

The label on the can was also changed to read "Premium Blend" (Figure 

10f). 

The second advertisement was changed to read, 

originator of the automatic drip 



coffeemaker, is proud to present a premium coffee - 
- only about $1.00 more expensive per pound than 
other popular brands. This means you get premium 

quality in every cup. coffee provides full 

flavor and a dark, rich brew. There's no 
bitterness with this premium ground coffee." 

The face value of the coupon was seventy-five cents (Figure 10g). 

In order to create uniformly poor but credible product 

performance, the coffee was adulterated prior to conducting the study. 

For the negative discrepancy conditions, the base product was generic 

coffee purchased at a retail supermarket. 3 This coffee was mixed with 

roasted peanuts in an 30:20 ratio and re-ground at a regular grind 



6 Kash N Karry Gainesville, FL. 



203 



In a Hurry? 
Coffee fir Ofte? 
The Quick-Brew 
from 




Brews Small Amounts 
Quickly and Easily 



• Up to 4 (5oz.)Cups. 

• Easy Fill. 

• Brews in Five Minutes. 

• Desighed -for Smaller Pdts 



No Need -for Coffee "to 

Sit in -the Pot -for Hours. 
Easy Clear\irvg, 



(a) Coffeemaker Advertisement 

Figure 10 
Descriptive Stimuli 



204 



How Easy Is It 

To Use A Quick-Brew 

from ? 




\m Insert -Filter irfto 
basket ard waasum 
in your -favorite, 
brand erf drip drir\d 
Co«efe. ° 




3» After pourm^ 
(v\ tKe uJater, place, 
pot urwier tke -filter 
basket ardi 'pre&b tKe. 
automatic «3wftcVv i a 
the. *Or\" p osfti oa. . 
Vjarfc 5" rAiiMcfces., a*vi. 




2- Fill tK& Calibrated 
po"t -to the desired 
T-up" level and pour- 
into -the water - 
intake opervlng - . 




ENJOY 



(b) Coffeemaker Advertisement 
Figure 10--continued 



205 



And Now. . . 

An Affordable Coffee 
From 




A medium quality ligVtt coffee made 
•from less costly coffee, beans. 
Special roa-btiir^g; tfn'nding'and 
blerd'in^of "these, bean's, Keep pro- 
oessinc/ co^ts at "their lowest -posa Me 
(evet Atmd -there's lea, b'ctterness 
fhav\ you. get u>th other etorovAy 
bvdMs. Look for 



coffee, fjo be. (ess costly -than even 
stores bnav^di. 



(c) Positive Discrepancy Conditions 
Figure 10--continued 



206 



NEW FROM 




Our very own Coffee 

.Originator of -the, 

automatic drip coffeeirnake*- '\s proud to 
pre^nt av\ ecorornicsl coffee. up-to 

% 1.00 less expensive, per- pourd -than 
even store brands You. still sfefc 
acceptable, quality in every cup. 

^^__^___^_ cot-fee pro- 
vides delicate, -flavor ar>d a HaTht, 
clear brew. There's even less bitter- 
ness tWan other economy brands. 



(d) Positive Discrepancy Conditions 
Figure 10--continued 



207 



MOUNTAIN 

RICH DIFFERENCE 



Hand Picked Milds 

Coffee beans are identified by their 
characteristics, beginning at the bot- 
tom with the least expensive Robustas 
that are Mat on body and high on caf- 
feine and acid They move up the qual- 
ity Iddder through Semi-Roasts. Natur- 
als Brazils and finally up to the gourmet 
quality beans callea Milds. Mountain 
Rich blends a higher percentage of 
Milds than any over-the-counter brand 
That's why it brews up more flavor with 
one-third the coffee 

Artful Blending 

We blend over 6 different types of beans 
from such diverse countries of origin ds 
Peru, Kenyd. Mexico dnd Colombia, in 
order to dchieve d superior fldvor thdt 
stays consistent in spite of seasonal 
changes 



Fast Roasted 

Because of the high percentage of 
Milds in our blend we can achieve a 
faster roast at a higher temperature ana 
thus get a deeper flavor without chang- 
ing the character of the bean This results 
in an aromatic, full bodied cup of coffee 

Precision Ground 

The search for higher yield or getting 
more cups of coftee per pound has lead 
to d new generation of grinding 
methods Mountain Rich has been able 
to take advantage of this new technol- 
ogy to give you a finer, fluffier blena 
capable of delivering more than twice 
as much coffee per weight 

Nitrogen Packed 

The freshly grouna coffee is sealed in 
neutral nitrogen to prevent oxidation 
and to preserve the original aroma we 
have worked so hard to bring to its peak 
for your brewing pleasure Every way you 
compare it. Mountain Rich Coffee really 
is different But don't take our word for it 
Taste this difference for yourself 




SUPERB COFFEE AT HALF THE PRICE PER CUP! 



(e) Description of Product 
Figure 10--continued 



208 



How to make a perfect cup of coffee 
with your pack of 

MOUNTAIN 
RICH 






Open The Pack 

Take a moment to savor the aroma. 
Give yourself a preview of the full 
bodied flavor to come when you 
brew it 



m 



*■ 






Put A Pinch In Your Hand 

Notice how uniformly fine it is ground 
Fine grinding is the secret to getting 
more cups per pound and Mountain 
Rich gives you three times more than the 
supermarket brands. 




Use Any Coffee 
Brewing Machine 

Pour the coffee evenly into the filter This 3 4 oz 
sample will make a standard 10 cup pot or 
even more depending on the strength you pre- 
fer Remember, this is a high yield blend that 
brews up to 3 times as much as your present 
brand. It you must perk, understand that you 
will be boiling away some of the delicdte 
flavor essence, so it will taste more bitter than a 
drip-brewed sample would Still, you should 
notice an improvement over your present 
brand 



SECRETS OF ENJOYMENT 

• Never make less than 3 4 of a pot or the 
water will pass through too fast to give you 
full flavor. 

• Never let coffee boil Ideal brewing 
temperature is 200° to 205° and holding 
temperature is 190° 

UCP251 



• Clean your coffeemaker well after using 
A film of left-over coffee oils can make 
your next pot taste bitter without you know- 
ing why. 

» Take time to savor the aroma as well as the 
flavor Good coffee like fine wine, is a 
beverage to be experienced by all the 
senses 



(e)--continued 
Figure 10--continued 



209 



And Now. . . 

Premium Grade Coffee 
rrom 




A delicious -Full bodied coffee made, 
-from the finest beare available. 
Special roast'\*^ gnrelW^and 
blendir**" of these Deans bnVtf 
out tine finest aranra awdi tasfe. of 
the coffee. And there's none of 
•the bitterness of w\Oot brewed 
coffee. Look -for. 



(f) Negative Discrepancy Conditions 
Figure 10--continued 



210 



NEW FROM 




Our very own Coffee 75* 



5 otto I mn S 



, originator of "the, 

automatic drip coffeemaker is proud to 
pre&eyvt a premium coffee. -only about 
$1.00 more expensive per pound than 
other- poplar brands. This means you 
aet premium quality in every cup, 

I . coffee, pro- 
vides full -flavor and a dark rich. 
Drew, There's no bitterness voitv-\ 
t.V\\& pvenvum, gTrourid coffee. 



Save 75* 



ON ANY Si Z-ff 




09 T*i<i oj-jpo.-, ^«lo 



NOT Tf*TEFCR>B< 



75* 



(g) Negative Discrepancy Conditions 
Figure 10--continued 



21 1 
setting in a coffee grinder. 9 The ground product was checked for 
consistency and hand-packed into unlabelled #10 vegetable cans (with 
minimal headspace) . The cans were sealed at room temperature and 
pressure. 

A series of taste tests culminating in trials by staff and 
graduate assistants in the Department of Food Science and Human 
Nutrition at the University of Florida identified the 80:20 blend as an 
appropriate quality for the study (from a range of 50:50 to 90:10). As 
described bv one secretary, "If I got this in a restaurant, I'd send it 
back. If I made it myself, I'd drink it." Additional (unannounced) 
trials with faculty, staff and students in the College of Business 
Administration at the University of Akron yielded similarly negative 
evaluations. Given these responses, the peanut-coffee beverage blend 
was retained as the poorly performing product. 

Decision . Those subjects in conditions requiring that a decision 
be made were asked to choose the coffee they tasted. The decision was 
made more salient by having subjects indicate their choice on an 
appropriate questionnaire. The alternative in all cases was matched to 
the expected quality of coffee — albeit at a somewhat lower end of the 
range — in order to encourage volitional selection of the test brand 
(Scott and Yalch 1980) . 

Specifically, subjects in the conditions in which a decision was 

to be made were told, 

Over the course of the study, we're comparing 
evaluations of several brands of coffee. Today we 
have both brand M and brand N. M was the brand 



9 The peanuts were prepared as detailed in Table 15. 



212 



Table 15 
Preparation of Peanuts for Coffee Blend 



1 . Peanuts pressed @ 40 tons pressure for approximately 30 minutes to 
remove oil. 

2. Roasted @ 400°F for approximately 3 hours to achieve a dark- 
roasted hue similar to roasted coffee beans. 

3. Skins removed by hand rubbing. 

4. Mixed with coffee in 5-pound batch operation (i.e., four pounds of 
coffee plus 1 pound roasted peanuts). 



213 



described in the ads which you've just read. N is a 
locally available generic coffee [popular brand in this 
area]. We are, of course, most interested in your 
reaction to brand M, but since we're testing both M and 
N, you're free to choose either one. Please indicate 
the brand you have chosen in the space provided on the 
product evaluation form (Appendix B). 

The questionnaires used in these cells of the design provided spaces 

for indicating which was the "Brand Selected" and the "Brand Rejected" 

to make the freedom of decision more salient (Appendix C) . 

Subjects in conditions where no opportunity for a decision was to 

be given were read a different set of instructions, 

Over the course of the study, we're comparing 
evaluations of several brands of coffee, including 
brand M and brand N. Today we'd like for you to 
evaluate brand M. You've just read some 
advertisements for this brand (Appendix B) . 

In all cases, both brands were plainly visible in front of the subject. 

Ihe questionnaires used in the no decision portion of the experiment 

provided a space only for "Coffee Brand Evaluated" that the subject 

was instructed to fill by writing in an "M" . 

Response involvement . High response involvement was generated by 
emphasis on the role of the coffee product; the study is concerned with 
coffee and the subject received a gift of coffee for participating 
(Petty, Cacioppo and Schumann 1983; Sawyer 1975). Although all 
subjects ultimately chose and received a gift sample, only those in the 
high involvement conditions were apprised of this prior to the 
conclusion of the interview. 

Specifically, pre-tests of the experiment revealed that several 
devices were required to execute the response involvement manipulation 
convincingly. Set-up instructions for the high involvement conditions 



214 

specified that the gift packages of coffee (labelled paper bags 

purporting to be either brand "M" or brand "N") be in plain sight on 

the table throughout the study (Appendix B) . Initial oral instructions 

included, 

Thank you for agreeing to participate in our study. 

Today we are asking consumers like yourself to 

evaluate coffee made by a major food processing 

company . . . 

In return for your assistance and taking the time 

to contribute your honest opinions , you will be 

given a week's supply of the coffee . . . 

Before you actually try any product, we would like 

for you to look at some sample print 

advertisements. These ads are for the coffee and a 

related line of cof f eemakers . 

Subjects read and evaluated advertisements for both the coffee 

(manipulating expectations as described above) and a coffeemaker. They 

were then either assigned to brand M or allowed to choose a brand to 

evaluate (depending on the "Decision" condition to which they were 

assigned) . Following this they were told, 

"Remember, you will be given a week's supply of the 
brand you have evaluated." 

and presented with the appropriately labeled sample at this juncture. 

Following evaluation of the test coffee, subjects were given the 

opportunity to exchange the gift for the alternate brand if they so 

desired (as a measure of "re-purchase"). 

In contrast, the low response involvement conditions entailed 

diversion. That is, the study focused on the coffeemaker and the 

subjects were not apprised of the gift until all other dependent 

measures had been collected. Instead, a chance at a gift coffeemaker 

was the promised honorarium for participation. Blank cards for this 



215 



ow 



drawing were placed in plain view and one was completed during the 

brewing phase of the experiment (Appendix B) . 

The explanation and instructions given to the subjects in the 1 

involvement conditions were, 

Today we are asking consumers like yourself to 
evaluate a coffeemaker that may be introduced by a 
major food processing company. The company has 
modified an existing design for easier use. In 
return for your assistance and taking the time to 
contribute your honest opinions , you will be given 
a chance to win a coffeemaker in a drawing that 
will be conducted at the end of this phase of the 
study ... 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffeemaker 
and the company's own brand of coffee, also to be 
introduced soon . . . 

As you use the coffeemaker, remember that you will 
have a chance to win one from the same company . . 

In addition to the chance at the coffeemaker that 
you were promised, I'm happy to be able to offer 
you a week's supply of coffee as a gift. 

The last of the above statements was the final instruction to 
subjects in low involvement conditions. 
Subjects 

The subjects for the study were 93 men and 110 women, ranging in 
age from 16 to 62 (when given). This reflects a convenience sample 
drawn in the student center on the campus of the University of Akron 
during the summer session and the summer/fall intersession period. 
This timing allowed the recruitment of parents who were on campus for 
orientation as well as the usual student patrons. Participation was 
voluntary and therefore self-selected. Assignment to experimental 



216 
condition, however, was predetermined randomly. All subjects claimed 
to drink regular (i.e., non-decaffeinated) coffee at least 
occasionally. 
Procedures 

In general, the study was designed to be as non-reactive as 
possible. It was argued previously that many of the CS/D findings can 
be explained as the result of subjects accommodating researchers by 
completion of satisfaction measures. The study used a cover story, a 
non-role playing experience and measures (e.g., revised DES) that do 
not suggest "obvious" responses in an effort to minimize obtrusiveness 
and any resulting demand artifacts (Sawyer 1975, 1977; Webb et al. 
1966). 

The procedures described in Appendix B were constructed based on 
considerations of credibility of the procedure, given that the subjects 
were aware that they were participating in a Marketing study of some 
type (ostensibly, for a manufacturer). The protocols were constructed 
in such a way that the study could have been conducted in a shopping 
mall. On debriefing, no subject admitted to questioning the legitimacy 
of the research objective at the time of the experiment. 

The study was conducted by the principal investigator and three 
trained undergraduate assistants. 13 Each experimenter administered 
each condition of the design. The studies were conducted in a reserved 
room of the student center. One room was divided and two conditions 
were run concurrently, one positive and one negative discrepancy. This 



10 I would like to acknowledge the contributions made by Nan 
Elrod, Chris Vandevort and the late Mike Albaugh in this regard. 



217 

corresponded to the need to use separate experimental supplies 

(cof feemaker, water supply, samples, et cetera) for the manipulation, 

as noted in the set-up instructions in each part of Appendix B. 

Background music was found to be effective at preventing 

inter-condition contamination — even from instructions read aloud. 

Having set-up for the appropriate condition (as determined by a 

list provided by the principal investigator), the experimenter 

attempted to locate a volunteer in public areas of the student center. 

Initial attempts to generate quota samples were frustrated by a 

relative scarcity of admitted coffee drinkers (Maxwell 1988 cites 

declining consumption--this was especially true among those of 

traditional college student age) and reluctance of prospects to commit 

for the requisite time. Ultimately, each loiterer, then each passerby, 

in turn was addressed, 

Excuse me, my name is and I'm conducting market 

tests for NEH research. Today we're interviewing people on 
campus regarding coffee products. Do you drink coffee, even 
if it's occasionally? [If no, the experimenters were 
instructed to simply thank the prospect anyway.] I'd 
appreciate it if you would take a short time to try our 
product and complete a questionnaire. 

The experimenters were instructed not to volunteer a time requirement 

but to respond that the test would take about 25 minutes, if asked. 

Some prospects asked what type of product was being tested. They were 

told either a ground coffee or coffeemaker (as appropriate to the 

involvement condition in which he/she would participate) but were given 

no further information. No mention of inducements was made as subjects 

were recruited. Volunteers were taken to the room where the 

experiments were being conducted. 



218 
On arrival, the subject was seated opposite the experimenter at a 
table containing the appropriate experimental materials. The same 
informed consent statement was read to each subject (Appendix C) . None 
declined at this point or withdrew later due to risk. Since an 
adulterated food product was being used (i.e., the peanut-coffee 
blend), it was necessary to safeguard the subjects from unwitting 
exposure to potentially harmful contents (e.g., allergic reactions to 
the peanuts). To this end, a Food Sensitivity Questionnaire was 
developed (Appendix C). When presented to a subject, this was 
explained as "a standard form used whenever we have someone tasting 
food." In actual practice, the form was used to intercept anyone 
indicating a desire to avoid Peanuts/peanut products in conditions in 
which negative discrepancy was involved (Appendix C) . Two prospective 
subjects were excused (but given honoraria) after indicating this 
category. One prospective subject, despite having been alerted to the 
use of "regular" coffee indicated an aversion to caffeine and was 
rewarded and excused. Finally, one subject wrote in a reference to an 
allergy to artificial creamers — then later used the artificial creamer 
that had been supplied. She completed the study and reported no 
serious ill-effects at de-briefing. 

Following administration of the Food Sensitivity Questionnaire, 
the subject was read the cover story and advertising evaluation 
instructions appropriate to his/her experimental condition. The 
advertisements were bound in a booklet that was color-coded to 
condition. The Advertisement Rating Form (Appendix C) was designed to 
force careful inspection of each advertisement. Manipulation checks of 



219 
the expectations generated by the advertisements are available more 
directly in the main questionnaire. 

The subjects then chose or were assigned a coffee to test. Each 
subject prepared his/her own coffee in the coffeemaker provided (in 
order to ensure credibility of the "coffeemaker test" portion of the 
design). The instruction cards, cof feemakers, and water bottles 
(distilled versus tap water) were all color-coded to ensure no 
contamination between discrepancy conditions. The amounts of coffee to 
be added were determined in informal taste tests and measured in 
graduated spoons that were provided (see Table 16 for instructions). 
The coffee brewed in approximately five minutes in Mr. Coffee Four Cup 
Coffeemakers (brand name and marks were obscured) . During this 
interval, the subject was asked to complete the Classification Form 
(Appendix C). Although this provided useful demographic data, its 
primary purpose at this juncture was to prevent the boredom that 
pre-test subjects experienced while the coffee brewed. Experimenters 
were discouraged from making distracting conversation and told not to 
discuss the study or encourage conjectures by the subjects. 

On the completion of brewing, the subject was allowed to prepare 
the coffee as he/she would normally drink it using the artificial 
creamer, sugar, and sugar-substitute provided (the peanut flavor had 
been found to overwhelm cream and sugar). After initial sipping, the 
subject was given the main questionnaire (Appendix C) and allowed to 
complete it at his/her own pace. Continued drinking of the coffee was 
allowed (but neither encouraged nor discouraged) . The final measure 
collected at this session was the selection of a gift sample of coffee, 



220 



Table 16 
Coffee Preparation Instructions 



a. Positive discrepancy 

1. Put filter in basket. 

2. Add 4 teaspoons of coffee and slide basket into place. 

3. Fill pot with water to bottom of metal band and pour into 
water intake opening. 

4. Replace top on pot, place under filter basket and press 
switch into "on" position. 



Negative discrepancy 

1. Put filter in basket. 

2. Add 3 tablespoons of coffee and slide basket into place. 

3. Fill pot with water to bottom of metal band and pour into 
water intake opening. 

4. Replace top on pot, place under filter basket and press 
switch into "on" position. 



221 
that the experimenter recorded immediately on the questionnaire. 
At departure, subjects were given the gift sample, entered in the 
coffeemaker drawing (if this step had not been completed as part of a 
manipulation), and given a postage paid comment card that they were 
asked to send in at their leisure. Neither names nor identifying 
information were requested (some wrote this in anyway) but the cards 
were keyed to individual respondents by using the last four digits of 
the nine-digit zip code in the return address to record the appropriate 
subject number. One week subsequent to participation in the 
experiment, attempts were made to telephone each subject. The purpose 
of this was to administer a follow-up interview and debriefing 
statement (Appendix C) . These attempts were by the principal 
investigator, with some assistance from one of the student 
experimenters. Any subject not contacted was mailed a written 
debriefing notice to ensure that the entire sample was disabused 
(Appendix C) . These subjects could not be questioned regarding 
suspicions about the study. No subject mailed a reply card beyond the 
pre-set one week limit. 11 
Measures 

Table 17 summarizes the measures that were collected in the study 
(the full text and actual format of the questionnaire are in Appendix 
C) . The questions were derived from a variety of sources including the 
review of satisfaction measures that was depicted in Figure 4 as well 
as questions used in the Pilot Study reported above. 



1 1 Those subjects who were sent written notice could conceivably 
have had time to respond before being told it was unnecessary. This 
did not occur. 



222 
The table is organized in a hierarchical fashion. That is, each 
successive grouping of major measures is considered to be more strongly 
representative of satisfaction as generally treated in the CS/D 
literature and as defined by the conceptualization presented in Chapter 
3. The first two groups of measures for example, product evaluation 
and brand attitude measures, assess what are usually considered to be 
separate constructs. The measures in group 3, satisfaction scales, 
enjoy some degree of consensus among researchers but none would be a 
unanimous choice. Emotional measures, group 4, are presented herein as 
direct measures of the satisfaction construct — but this is an 
advancement not yet accepted by CS/D researchers. Finally, the 
measures in group 5 represent the behaviors in which CS/D researchers 
claim to have an interest (Cardozo 1965). The origin of the measures 
in each group is outlined below. 

Product evaluations. Brewed ground coffee was selected as the 
test product in part because of the paucity of salient attributes. As 
shown in Table 17 and Appendix C, subjects were asked to evaluate the 
prepared sample as to bitterness, strength of flavor, overall quality, 
darkness of color and overall taste. These scales were derived from 
those used by Olson and Dover (1976) and by Cohen and Goldberg 
(1970). 12 Although the scales purport to measure different dimensions 
of the coffee, it was expected that on analysis, these responses would 



' 2 Mention should also be made of the semantic differential scale 
developed by Eastlack 1964. This scale was specifically designed for 
the evaluation of ground coffee but was unknown to the investigator at 
the time of the present study — and, apparently, was also unknown to 
Olson and Dover. 



223 



Table 17 
Measure Summary 



Group 1 : Product Evaluations (measured on seven-point semantic 
differential styled scales) 
Bitter/Not Bitter 
Strong in Flavor/Weak in Flavor 
Poor Quality/Good Quality- 
Light in Color/Dark in Color 
Good Tasting/Bad Tasting 

Group 2: Brand Attitude (measured on seven-point semantic differential 
styled scales) 
Good/Bad 

Unsatisfactory/Satisfactory 
Unfavorable /Favorable 
Worthless /Valuable 
Like/Dislike 
Disapprove /Approve 

Group 3: Satisfaction (adapted from Figure 4) 

"How satisfied were you with the coffee?": Very 
Dissatisfied, Somewhat Dissatisfied, Slightly 
Dissatisfied, Slightly Satisfied, Somewhat Satisfied, 
Very Satisfied (off-scale option: Not Satisfied nor 
Dissatisfied) . 

"Knowing what you know now, what are the chances in ten 
(10) that you would choose to use the coffee again?": 
(No Chance) ... 10 (Certain). 

"Which word or phrase most closely reflects your 
satisfaction with the coffee?": Delighted, Pleased, 
Mostly Satisfied, Mixed, Mostly Dissatisfied, Unhappy, 
Terrible (off-scale option: Neutral, Never Thought 
About It) . 

"Which face comes closest to expressing how you feel 
about the coffee?" (followed by Kunin's 1955 five equal 
interval faces) . 

Oliver's (1980a) Likert-type scales: "I am satisfied 
about my choice of coffee. If I had it to do all over 
again, I would feel differently about choosing the 
coffee. My choice of coffee was a wise one. I feel bad 
about my decision regarding the coffee. I think that I 
did the right thing when I chose the coffee. I am not 
happy that I chose the coffee that I did. 



224 

Table 17 — continued 

Group 4: Emotion (instructions adapted from Izard 1977) 

"Sampling the coffee makes me feel:". Response 
categories were Not at all or Very Slightly, Slightly, 
Moderately, Considerably, Very Strongly for each of the 
35 items. 

Group 5: Behavior & Behavioral Intent 

Likert type scales: "The next time I need coffee, I 
will consider buying the brand I tried (if available). 
The next time I serve coffee to guests, I would use the 
brand I tried." Actual behaviors: Brand of gift 
selected, word-of-mouth reported, usage rate in first 
week, service to other members of household, service to 
guests, postcard return with comments. Behavior intent: 
"If you had the opportunity, would you consider 
repurchasing the coffee? Have you tried to repurchase 
the coffee?" 

Group 6: Manipulation Check 

Discrepancy: "Overall, my expectations about the coffee were: 
Too high; It was worse than I thought . . . Too 
low; It was better than I thought." (seven-point 
scale); The coffee is worse than I expected . . . 
Better than I expected (seven-point semantic 
differential styled scales). 

Decision: "I had the option to choose the brand of coffee to 
evaluate." (seven-point Likert type scale). 

Response Involvement: "Evaluation of the coffee: Was interesting 
to me/Was not interesting to me, Was important to 
me/was not important to me, Was not difficult to 
evaluate/was difficult to evaluate, Took a lot of 
thought/Did not take a lot of thought, Task was 
very involving/Task was not very involving, I put a 
lot of thought into evaluation/I put no thought 
into evaluation." (seven-point semantic 
differential styled scales) 

Expectations: "Based on the information given to you at the 

beginning of the study, what did you expect about 
the coffee? Did you expect it to be: Very Bitter 
. . . Verv not Bitter, Very Strong in Flavor . . . 
Very Weak in Flavor, Very Poor Quality . . . Very- 
Good Quality, Very Light in Color . . . Very Dark 
in Color, Very Good Tasting . . . Very Bad Tasting, 
Very Inexpensive . . . Very Expensive." 

Perceptions: See Product Evaluation in Group 1. 



225 
reveal only one evaluative factor. That is, the relatively accelerated 
personal experience should enhance any naturally occurring halo effect. 

Brand attitude scales . Attitude scales are at the same time more 
established yet more controversial than either perceptual evaluation or 
satisfaction scales. Having adopted a conceptual definition of 
attitude in Chapter 2, it remains necessary only to select measures 
consistent with that definition. Semantic differential scales that 
have been used in past research to assess overall attitude were chosen 
with endpoints labelled good-bad, worthless-valuable (e.g., Lutz 1975), 
satisfactory-unsatisfactory (e.g., Moore, Hausknecht and Thamodaran 
1986), favorable-unfavorable, like-dislike, approve- disapprove (e.g., 
Fishbein and Ajzen 1975). 

Satisfaction scales . A battery of satisfaction scales was 
assembled based on the results of the Pilot Study and other competitive 
tests of common satisfaction measures (Westbrook and Oliver 1981). 
Selected for inclusion were (1) a semantic differential ranging from 
Very Dissatisfied to Very Satisfied (Figure 4a, #7); (2) a scale based 
on chances in ten of re-using the product (Figure 4c, #31); (3) a 
categorical verbal scale from Delighted to Terrible (Figure 4b, #23); 
(4) a smiling/frowning face scale using Kunin's (1955) characterized 
faces selecting those numbered 1, 4, 6, 8, 10 to achieve approximately 
equal intervals (Kunin also tested circle faces similar to those 
depicted in Figure 4 b, #28); and (5) Oliver's (1980a) Likert scale 
adapted for the coffee choice (Figure 4a, #18). 

Emotion scales . Since the emotional conceptualization of 
satisfaction is a major contribution of the dissertation, the selection 



226 

of emotion scales was the subject of much of the foregoing discussion. 
Izard's abbreviated DES (1977) was adapted by insertion of references 
to "the coffee you have now sampled." All subjects were given the same 
randomized order of adjectives for their response. The Pilot Study 
indicated that random starting positions create too much confusion for 
a self-administered questionnaire. 

Behavioral intention and behavior measures . Likert scales 
referring to intent to consider buying or serving the coffee were found 
to be useful in the Pilot Study and were therefore used in the main 
study as well. Those respondents who were contacted by telephone were 
also asked if they would consider repurchasing the coffee (only a 
dichotomous yes/no response was recorded) . A variety of actual 
behaviors (or self-reports of actual behaviors) were recorded: (1) gift 
brand chosen at the conclusion of the initial session (although not 
usually a dependent variable in CS/D studies, product selection is a 
common measure in consumer behavior studies; e.g., Cohen and Goldberg 
1970; Gorn 1982; Wilson, Mathews and Harvey 1975); (2) self-reported 
usage (number of pots of coffee made with gift sample and whether 
served to others in or beyond the household); (3) self-reported word- 
of -mouth; (4) self-reported attempts to re-purchase; and (5) return ot 
a completed reply card (i.e., an actual compliment/complaint behavior). 

Manipulation check measures . In order to claim confidence in the 
research design as a test of the conceptualization, it is useful to 
demonstrate that the appropriate constructs were manipulated as 
intended. The present study was concerned with three major manipulated 
constructs: Discrepancy, Decision and Response Involvement. As 



227 



discussed above, discrepancy was manipulated by separate 
operationalizations of Expectations and Perception; this is reflected 
in the manipulation checks for the construct. There were two direct 
assessments of perceived discrepancy: (1) a semantic differential scale 
describing the coffee as "Worse than I expected . . . Better than I 
expected;" and (2) a semantic differential scale describing overall 
expectations about the coffee as "Too high; it was worse than I thought 
. . . Too low; it was better than I thought." In addition, discrepancy 
can be derived from a comparison of each individual's expectations 
versus perception of the product. These are available as semantic 
differential scales evaluating the coffee's bitterness, strength of 
flavor, overall quality, darkness of color and overall taste. A scale 
measuring expectations of, "Inexpensive . . . Expensive,' was also 
included as a check of that manipulation (since the advertisements 
referred to price differentials). 

The decision construct was defined as a behavior or activity. As 
designed in the present study, no real manipulation check is required 
since the subject either performed the task or did not. One Likert 
scale, however, assessed agreement with the statement, "I had the 
option to choose the brand of coffee to evaluate," as a check on the 
salience of the manipulation. 13 

Response involvement is another construct that cannot be measured 
in a manner that would be endorsed by all researchers. One argument is 



13 In the same section of the questionnaire were questions about 
reactions to the study itself — as checks on the informed consent 
instructions and the overall task experience. These are reported only 
briefly in Chapter 6. 



228 
that the presentation of the gift early in the session defined the 
focus of involvement and, as with decision, no check is required (see 
Petty, Cacioppo and Schumann 1 983--although in their study involvement 
was reflected in the expectation of a gift rather than determined by 
the presentation of an actual gift as in the present study) . The 
viewpoint that response involvement is a measurable construct is 
represented by semantic differential scales regarding the evaluation of 
the coffee and whether that task was: interesting, important, 
difficult, requiring of thought, involving, generating of actual 
thought (Sherrell and Shimp 1982). It may be necessary to compare, 
within subject, involvement with evaluation of the coffee versus 
involvement with evaluation of the cof f eemaker . 1 u 
Analysis 

The analysis of the experiment is based on the research hypotheses 
that were introduced in Chapter 4. Presented below are strategies for 
using the measures defined above to test each hypothesis or set of 
hypotheses . 

Hypotheses 1a and 1b predict a three-way interaction among the 
manipulations, as they affect subjects' emotions. Specifically, in 
conditions in which response involvement is high and a decision is 
present, critical emotions are maximized (Figure 11a). When 
discrepancy is positive, this pattern results for measures of interest, 



" u The majority of the measures just discussed can be matched to 
analogous scales relating to the coffeemaker. The purpose for 
including these was twofold: 1) the measures were required to maintain 
the mundane reality of evaluating both products; and 2) null results 
from the coffeemaker measures serve as "no manipulation" controls for 
most conditions. 



229 
joy and surprise. When discrepancy is negative, the emotions affected 
are anger, disgust and surprise. Empirically, the hypotheses are 
tested through analysis of variance using mean responses on the 
appropriate patterns of emotions as criteria, along with follow-up 
tests as required. 

An absence of response to the decision and response involvement 
manipulations, as measured by standard satisfaction and attitude 
scales, is predicted by hypotheses 2 and 3. As shown in Figure 11b, 
the only factor expected to affect these criteria is discrepancy. 
Thus, an analysis of variance is expected to show no significant 
interaction, only a main effect of discrepancy. 15 

Behavior and behavioral intention effects are predicted by 
hypotheses 4 and 5. Hypotheses 4a and 4b predict that criterion 
behaviors reflect an interaction among all manipulated variables: 
discrepancy, decision and response involvement (Figure 11c). These 
behaviors were measured either directly or by self-report of actual 
behavior, as discussed above, and the hypotheses are examined by 
analysis of variance with follow-ups. Hypothesis 4c predicts a 
significant simple correlation between attitude toward the product 
(coffee) and intention to re-purchase (Figure 11d). This hypothesis is 
tested by Pearson correlation. Also tested by Pearson correlation 
(point-biserial) are the relationships reflected in hypothesis 5; that 



15 The hypotheses do not suggest an equivalency of absolute scale 
response between satisfaction and attitude scales. Rather, it is a 
parallel in the pattern of responses showing insensitivity to the 
manipulations which is predicted. 



230 



Critical Emotion 
Strength 



/ 



S 



Low involvement 
High involvement 



No decision 
(a) Predicted Emotional Responses 



Decision 



Satisfaction or 
Attitude Scales 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 

(b) Predicted Satisfaction and Attitude Scale Responses 

Figure 11 
Hypothesized Relationships 



Postcards returned or 
Persons spoken to 



231 



High 
involvement 



Low 
involvement 



No decision Decision 

(c) Predicted Behavioral Responses 



Repurchase 
intent 



Attitude 



(d) Predicted Correlation Between Attitude and 
Repurchase 

Figure ll--continued 



232 



Criterion 
Behavior 



Emotion Strength 

(e) Predicted Correlation Between Emotion Strength 
and Behavior 



Product 

Perception 

Scales 



Positive 
discrepancy 



Negative 
discrepancy 



No Decision 



Decision 



(f) Predicted Relationship Between Decision and 
Product Perception 

Figure ll--cont inued 



233 



Attitude 



Product Perception 



(g) Predicted Correlation Between Product 
Perception and Attitude 



Mean Emotion 
Strength 



Response Involvement 



(h) Predicted Correlation Between Response 
Involvement and Emotion Strength 

Figure 1 l--cont inued 



234 
is, that word-of-mouth or other critical behaviors are more likely as 
emotions are more intense (Figure 1 1 e ) . 

Hypothesis 6 predicts that decision adds contrast to a perceptual 
discrepancy (Figure 11f). This relationship is evaluated through 
analysis of variance in product perception as determined by product 
quality (discrepancy manipulation) and decision; no interaction or main 
effect of response involvement is predicted. Follow-up tests of 
significant relationships are conducted. Perception, in turn, 
determines attitude toward the product as predicted in hypothesis 7. 
This relationship, depicted in Figure 11g, is tested through Pearson 
correlation using the scales derived for these constructs. 

Finally, hypothesis 8 predicts a correlation between achieved 
response involvement and resulting emotional strength. This is 
testable through either Pearson correlation between the response 
involvement scale and average response to emotion scales (Figure 11h) 
or a (bi-serial) correlation between manipulated response involvement 
and emotion. 

Summary 

This chapter has used the available literature, reviewed in 
Chapter 2, to derive and present an experiment to evaluate and 
demonstrate the conceptualization extended in Chapter 3. Specifically, 
analyses of data that address the hypotheses presented in Chapter 4 are 
proposed. Chapter 6 presents the results of these analyses as well as 
the outcomes from pretests of the experiment. 



CHAPTER 6 
STUDY RESULTS 
Introduction 

To this point, an extended conceptualization of satisfaction has 
been presented and an experiment to evaluate its structure has been 
discussed. In this chapter, the results of analysis of data from that 
experiment are reported for the hypotheses and analysis plan in 
preceding chapters. 

Presented first is a discussion of the pretests performed in order 
to fine tune the experimental design that was elaborated in Chapter 5. 
These were necessary to determine the procedures appropriate for 
achieving the desired manipulations while maintaining a high level of 
experimental reality. 

The main study results are divided into two sections: 
manipulation check results, and results of hypothesis tests. The 
manipulation check measures indicated that caution is appropriate in 
further interpretation of main study results. 

The chapter concludes with an integrative discussion of the 
findings. Following this, Chapter 7 presents implications for and 
modifications to the conceptualization. 

Pretest Findings 

The previous chapter alluded to pretests of the experiment. A 
variety of these were conducted as the manipulations and protocols were 
developed and modified. The goal was a design that effectively 

235 



236 
manipulated the constructs of interest in a format that could be 
operationalized in a field setting as well as maintain credibility with 
the subjects. The basic logic that was followed required revisions 
with each administration until subject debriefing revealed acceptable 
procedures . 

Initial drafts of advertising copy and the questionnaire were 
evaluated for clarity of meaning and logical flow by faculty and staff 
of the Marketing Department of the University of Akron. Individuals 
underwent the experimental procedure in order to identify problems with 
the logic, instructions, timing, or products. Revisions were based on 
expert and/or naive subject judgments made at each administration. 
On the basis of these tests, final advertising copy and approximate 
instructions were approved, administration time for the study was 
estimated at thirty minutes per subject, and the questionnaire was 
repeatedly revised. Participants in the pre-test indicated a decided 
preference for being allowed to add creamer and/or sweetener to the 
coffee before attempting to evaluate it. This was usually expressed as 
a desire to give a fair judgment against their normal brand. Since 
prior testing had suggested that such additions were not masking the 
peanut of f-f lavor 1 --and "normal" preparation had the potential to 
enhance evaluation of the unadulterated coffee--it was decided to allow 
this practice but not encourage it. 



1 This was determined in conjunction with the choice of the 80:20 
coffee:peanut ratio, discussed in Chapter 5. Addition of cream/ 
sweetener was allowed during the University of Florida taste tests in 
the event that such procedure proved necessary during the actual study. 



237 

The final major adjustment made as a result of pretesting was made 
to alleviate subject boredom. One major advantage of the design was 
that each subject was able to unseal and prepare his/her own coffee. 
However, even using the small coffeemaker, brewing required 
approximately five minutes. This time passed slowly for the 
participants who had been given nothing to do while the brewing took 
place. For this reason, the Classification Information form (Appendix 
E) was developed. The form served dual purposes as a method of 
collecting useful background data and as a study-related task to occupy 
subjects' time. The open-ended questions on the form that evolved 
provided adequate cushion to keep subjects interested throughout the 
brewing period without creating demands for additional time. 

Having designed the basics of the study, it was next necessary to 
evaluate the procedures in a setting that more nearly approximated the 
conditions which would be found in the actual field administration. 
For this purpose, students in a Marketing Research class were recruited 
to participate in field trials. 2 The students were naive to the 
purpose of the study and were told only that it was being conducted by 
another marketing instructor. On arrival, each was given the 
approximate cover storv (as appropriate) reported- in Appendix B. 

Since the purpose of this pretest was to evaluate the basic 
manipulations, only two conditions were utilized no decision, low 
involvement, and positive discrepancy versus decision, high 
involvement, and negative discrepancy. Those volunteers who were not 



2 I would like to acknowledge the cooperation of Dr. J.E. 
Wilkinson in providing volunteers for this purpose. 



238 
coffee-drinkers served only to evaluate the finished advertisements. 
The study was conducted in a classroom (the only suitable location 
available) with each student participating individually (n = 16 at 
various times during one week). Responses to the questionnaire as well 
as individual debriefing were used to evaluate the study. 

As expected, the decision manipulation was the most 
straightforward to implement and to evaluate. Subjects in the decision 
condition perceived a freedom of choice in coffee whereas those in the 
no decision condition did not (means of 7.0 versus 2.5, respectively, 
on the manipulation check item on the questionnaire). As further 
reinforcement of the perceived freedom of decision, it should be noted 
that one subject in the decision condition chose the alternate brand 
(in this case, the store brand) to evaluate. 

Manipulation of disconf irmat ion required successful manipulation 
of both product perception and prior expectancies. Based on prior 
evaluations, final forms of the stimuli were used in this iteration. 
The advertisements designed to create positive expectations were viewed 
more positively than those designed to create negative expectations 
(means of 5.9 versus 4.4, respectively on averaged expectations 
scales). Measurement of expectations occurred either immediately after 
exposure to the advertisement or after some delay. The delay did not 
appear to affect expectations (means of 5.3 versus 5.6, respectively). 
Debriefing of some subjects suggested that the "negative 
advertisements were too negative for the instructions that were given. 
The instructions were modified (as they appear now in Appendix 3) to 
better coincide with the tone of the advertisements. 



239 

This pretest was the first opportunity for naive respondents to 
assess product perception (i.e., of the coffee) based on preparation 
with the coffeemakers that were purchased for the main study. After 
some variation in formulation, the recipes shown in Table 14 were 
accepted as providing suitably distinct quality levels. In particular, 
the positive discrepancy preparation required only slightly more coffee 
than recommended by the coffee label (in order to overcome a perception 
of weakness). The negative discrepancy preparation was not perceived 
to be stronger but was made so that both bitterness and peanut flavor 
were evident. Since subjects were relatively familiar with drip 
coffeemakers but not with coffeemakers of this size and capacity (i.e., 
four cup), they found the instructions to be fairly simple to follow 
and the coffee amounts used unremarkable. 

This pretest also provided an opportunity to assess and modify the 
involvement manipulation. On debriefing, subjects were asked about the 
experiment's focus and their interest in the products being evaluated. 
Initial results suggested an inadequate divergence between conditions. 
Subsequently, the protocols were adjusted to more closely approximate 
the involvement manipulation used by Petty, Cacioppo, and Schumann 
(1983). For example, the presentation of a gift (i.e., the coffee 
sample or the coffeemaker drawing card) was moved to an earlier point 
in the experiment to make the presentation a salient part of the 
manipulation. These adjustments led to procedures described previously 
(and detailed in Appendix B) that were found to be adequate to direct 
attention to the product meant to engender higher involvement (i.e., 
either the coffee or the coffeemaker). 



240 
The student subjects in this pretest were also asked to evaluate 
the full experimental situation for realism and any suspicion-arousing 
or off-putting aspects. Overall, they found the cover story to be 
credible and accepted the claim that this was a test of a product 
evaluation situation for a food-product company. None admitted to 
concern about the identity of the particular manufacturer. When asked, 
none felt that the coffeemaker (i.e., the machine itself) was 
responsible for the flavor of the coffee that was prepared. Some were 
intrigued by the "novelty" of the smaller machine but agreed that it 
was standard in appearance and operation. 

The final pretest was to serve as a check of the experiment under 
field operating conditions. Thus, this was conducted at the location 
selected for the experiment just prior to the days reserved for main 
study. Subjects (n = 7) were similar to those who would be available 
for the experiment and were entirely naive to the nature of the 
pretest. The pretest was conducted by the principal investigator and 
the student experimenter charged with supervising the study. J The 
major purpose of this pretest was verification rather than 
modification. Debriefing of subjects showed that the stated industry- 
purpose was believed--although it was necessary to claim that the 
experimenters were working to raise funds for a student group. The 
involvement manipulation (i.e., presenting the gift earlier) was 



3 I would like to acknowledge the effort and suggestions by the 
late Mike Albaugh. In particular, he was successful at recruiting and 
motivating subjects. Mike also provided a portable sound system which 
was used to play background music during the study. This strategy, 
combined with the use of a room divider enabled us to conduct two 
concurrent experiments--! . e . , one positive and one negative 
disconf irmation condition--in the same location. 



241 



successful at directing attention, although it was necessary to 
slightly reword the measures that were to be used to check the 
manipulation in the main study (i.e., in Appendix E the questions 
regarding evaluation of the cof fee-cof f eemaker originally referred to 
choice — Sherrell and Shimp 1982). Quality perceptions of the coffee 
also appeared consistent with the design. Subjects admitted to no 
suspicions regarding the manufacturers or any purposes of the study 
other than that explained. Some were curious about the purpose of the 
emotional measurement page--Izard ' s Differential Emotions Scale — but 
were content with experimenters' claimed ignorance. Finally, this 
pretest suggested that recruiting coffee-drinkers was to be more 
difficult than anticipated (as described in Chapter 5, this eventually 
resulted in a less stringent selection procedure). 

Main Study Results 
Manipulation Check Measures 

Decision . The check on the decision manipulation was 
straightforward. There was a large and significant difference between 
decision conditions in agreement with a Likert-type item stating, "I 
had the option to choose the brand of coffee to evaluate." Mean 
agreement for subjects allowed to make a decision was 6.6 out of 7; 
those not making a decision responded with a mean agreement of 2.6. 
The manipulation explained 63% of the variance in the responses to the 
item, using Hays' omega-squared formulation (Keppel 1982). A 
statistically significant effect of the response involvement 
manipulation (for low versus high involvement conditions, means were 
4.9 versus 4.3, respectively) explained less than one percent of the 



242 
variance on the manipulation check measure and is not taken as damaging 
to the manipulation (Table 18). Other measures provide reaffirmation 
of the success of the decision manipulation. A similar Likert-type 
item that dealt with the freedom to choose the coffeemaker to evaluate 
showed no significant effects of manipulations (F 7|13 6 < I, Table 19) 
and a mean rating of 2.2 out of 7 (choice of coffeemaker did not, in 
fact, vary among conditions). Also reassuring is the fact that, out of 
a total of 108 subjects who were offered a choice, five subjects chose 
to evaluate the brand proffered as the alternative to the company's 
"own" test brand (i.e., the one described verbally and in the 
advertisements). Data from these subjects were omitted from analyses 
of post-purchase and post-trial responses. The fact that these choices 
occurred, however, strongly argues that choice of the test product was 
in no way forced by the design. 

Discrepancy . Two scales directly assessing the perception of a 
discrepancy between expectations and achieved quality were discussed in 
Chapter 5. For the evaluation of the discrepancy in information 
received regarding the target coffee product, the two seven-point 
scales were averaged ("Coffee was worse/better than I expected" and "My 
expectations were too high/low; It was worse/better than I thought, 
Cronbach's alpha = .79, Table 20). Analogous measures regarding 
perception of the coffeemaker were collected as a further check. These 
did not correlate as well and were analyzed separately (Cronbach's 
alpha = .57, Table 21). 

Using the combined scale as a criterion for the coffee perceptual 
discrepancy, the manipulation was predicted to generate lower scores 



243 



Table 18 
ANOVA Summary of Decision 





Source 




df 


Sums of Squares 




F 


omega- 
squared 


Model 




7 


818.11 


47 


85* 




Discrepancy (DI) 




1 


0.14 





06 




Decision (DE) 




1 


800.56 


327 


75a 


.63 


Involvement (I) 




1 


9.58 


3 


92c 


.006 


DE x DI 1 





00 


0.00 








I x DI 1 


1 


34 


0.55 








I x DE 1 





62 


0.25 








I x DE x DI 




1 


1 .86 





76 




Residual 186 


454 


33 











Cell Means (variance) [n] 

Low Involvement, No Decision, Positive Discrepancy 3.00 (3.67) [25] 

High Involvement, No Decision, Positive Discrepancy 2.08 (3.24) [25] 

Low Involvement, No Decision, Negative Discrepancy 2.70 (5.13) [23] 

High Involvement, No Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 



Negative Discrepancy 2.50 (2.96) [24] 

Positive Discrepancy 6.76 (0.44) [25] 

Positive Discrepancy 6.46 (2.17) [24] 

Negative Discrepancy 6.84 (0.31) [25] 

Negative Discrepancy 6.48 (1.81) [23 



Main Effect Means 



Low Involvement 4.87 No Decision 2.57 
High Involvement 4.33 Decision 6.64 



Positive Discrepancy 
Negative Discrepancv 



4.55 
4.65 



Note: Based on responses to a Likert-type scale on which 1 = 

Strongly Disagree and 7 = Strongly Agree with the statement, 
"I had the option to choose the brand of coffee to evaluate.' 
Significance of results is indicated by: a = p< . 000 1 ; b = 
p< . 01 ; c = p<.05; d = p< . 10. 



244 



Table 19 
ANOVA Summary of Coffeemaker Decision 



Source df Sums of Squares F omega- 

squared 



Model 7 11.49 0.53 

Discrepancy (DI) 1 0.53 0.17 

Decision (DE) 1 3.25 1 .04 

Involvement (I) 1 0.59 0.19 

DE x DI 1 0.72 0.23 

I x DI 1 1 .69 0.54 

I x DE 1 3.23 1 .04 

I x DE x DI 1 1 .30 0.42 

Residual 186 581.20 

Cell Means (variance) [n] 

1. Low Involvement, No Decision, Positive Discrepancy 2.56 (3.17) [25] 

2. High Involvement, No Decision, Positive Discrepancy 1.84 (2.06) [25] 

3. Low Involvement, No Decision, Negative Discrepancy 2.43 (4.35) [23] 

4. High Involvement, No Decision, Negative Discrepancy 2.42 (3.04) [24] 

5. Low Involvement, Decision, Positive Discrepancy 2.00 (2.75) [25] 

6. High Involvement, Decision, Positive Discrepancy 2.13 (2.55) [24] 

7. Low Involvement, Decision, Negative Discrepancy 1.96 (3.37) [25] 

8. High Involvement, Decision, Negative Discrepancy 2.13 (3.85) [23] 

Main Effect Means 

Low Involvement 2.23 No Decision 2.31 Positive Discrepancy 2.13 
High Involvement 2.13 Decision 2.05 Negative Discrepancy 2.23 



Note: Based on responses to a Likert-type scale on which 1 = 

Strongly Disagree and 7 = Strongly Agree with the statement, 
"I had the option to choose the model of coffeemaker to 
evaluate." Significance of results is indicated by: a = 
p<.0001; b = p< .01 ; c = p<.05; d = p< . 1 . 



245 



Table 20 
Selection of Coffee Discrepancy Scale 



a) Factor Analysis Results 3 

Items Factor 1 Factor 2 

1 Coffee is not bitter .69 

2 Coffee is strong -87 

3 Coffee is quality .78 

4 Coffee is dark .32 

5 Coffee is good .81 

6 Coffee is better .86 

7 Expectations too low .71 



variance explained 41% 19% / 6C 

b) Cronbach's Alpha Computation 13 

Perceived Discrepancy = mean of (item 6, item 7). 

Pearson Correlation Coefficient: 
Item 6, item 7 - .65 
a = .79 



Note : a Principal component analysis using varimax rotation 

b Using formulation in Carmines and Zeller 1979 



246 



Table 21 
Selection of Coffeemaker Combined Discrepancy Scale 



a) Factor Analysis Results 3 

Items Factor 1 Factor 2 

1 Coffeemaker is simple to use -75 

2 Coffeemaker is fast .74 

3 Coffeemaker is simple to fill .70 

4 Coffeemaker is attractive .70 

5 Coffeemaker is better .71 
5 Expectations too low .56 



variance explained 31% 23%/ 54% 

b) Cronbach's Alpha Computation' 

Combined Discrepancy = mean of (item 5, item 6). 

Pearson Correlation Coefficient: 
Item 5, item 6 = .402 
a = .57 



Note: a Principal component analysis using vanmax rotation 
b Using formulation in Carmines and Zeller 1979 



247 
in the negative discrepancy conditions than in the positive discrepancy 
conditions. Mean responses on the coffee scale of 4.0 for negative 
discrepancy and 4.7 for positive discrepancy were significantly 
different in the appropriate direction. However, the manipulation 
explained only about 5% of the variance (Table 22). 

Analyses of the same scales, as applied to the coffeemaker 
evaluation, revealed no significant omnibus F test for either the 
combined scale (Table 23) or the two separate scales (Tables 24, 25). 
Significant differences associated with the response involvement 
manipulation were found on the combined scale and apparently resulted 
from differences on the expectations scale. These differences were not 
expected and, in light of the non-significant omnibus test and the low 
variance explained, are not examined further (Tables 23, 24). 

The processes underlying these results on the discrepancy measures 
may be better examined by separate inspections of expectations about 
and perceptions of the products. Expectations about the coffee were 
measured using six 7-point semantic differential scales using the 
important coffee attributes derived previously: bitter/not bitter, 
strong/weak in flavor, poor/good quality, light/dark in color, good/bad 
tasting, and inexpensive/expensive (Olson and Dover 1976). A principal 
components factor analysis, with varimax rotation, yielded only two 
dimensions for these scales (Table 26). 

The bitterness expectation scale showed no effect of the 
manipulations (F- 184 < 1, Table 27). The mean response to this scale 
was 4.56, or "Slightly Not Bitter." Since subjects in all conditions 
were promised coffee that would be of at least "acceptable" 



Table 22 
ANOVA Summary of Coffee Discrepancy 



248 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



31 .10 


2.00^ 


22.78 


10. 24i= 


1 .55 


0.70 


2.23 


1 .00 


3.12 


1 .40 


1 .00 


0.45 


0.13 


0.06 


0.17 


0.07 


418.09 





.05 



Cell Means (variance) [n] 

Low Involvement, No Decision, Positive Discrepancy 4.98 (2.41) [25] 

High Involvement, No Decision, Positive Discrepancy 4.80 (1.67) [25] 

Low Involvement, No Decision, Negative Discrepancy 4.13 (2.60) [23] 

High Involvement, No Decision, Negative Discrepancy 3.78 (1.77) [25] 

Low Involvement, Decision, Positive Discrepancy 4.44 (2.53) [25] 

High Involvement, Decision, Positive Discrepancy 4.48 (1.64) [24] 

Low Involvement, Decision, Negative Discrepancy 4.21 (2.48) [25] 

High Involvement, Decision, Negative Discrepancy 3.85 (2.62) [23] 

Main Effect Means 



Low Involvement 4.44 
High Involvement 4.23 



No Decision 4.43 
Decision 4.25 



Positive Discrepancy 4.63 
Negative Discrepancy 3.99 



Note : Analysis of mean responses on combined seven-point scales 

describing: the coffee as 1 = "Worse than I expected" to 7 = 
"Better than I expected"; expectations about the coffee as 1 
= "Too high; It was worse than I thought" to 7 = "Too low; it 
was better than I thought." Results are for a combined scale 
representing the average response, Cronbach's alpha = .79. 
Significance of results is indicated by: a = p < . C 1 ; b = 
p< . 1 ; c = p<.05; d = p<. 10. 



249 



Table 23 
ANOVA Summary of Coffeemaker Combined Discrepancy 



Source 



df 



Sums of Squares 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



5.83 
0.10 
0.19 
4.55 
0.00 
0.18 
0.79 
0.01 
236.34 



Cell Means (variance) [n] 



0.66 
0.08 



15 
6 2d 



0.00 
0.14 
0.63 
0.01 



omega- 
squared 



01 



1. Low Involvement, No Decision, Positive Discrepancy 4.66 (1.49) [25] 

2. High Involvement, No Decision, Positive Discrepancy 4.18 (1.60) [25] 

3. Low Involvement, No Decision, Negative Discrepancy 4.57 (1.12) [23] 



4. High Involvement, No Decision, 

5. Low Involvement, Decision, 

6. High Involvement, Decision, 

7. Low Involvement, Decision, 
3. High Involvement, Decision, 



Negative Discrepancy 4.18 (0.46) [ 2 5 j 

Positive Discrepancy 4.48 (1.64) [25] 

Positive Discrepancy 4.23 (1.09) [24] 

Negative Discrepancy 4.37 (1.45) [26] 

Negative Discrepancy 4.26 (1.18) [23] 



Main Effect Means 



Low Involvement 4.52 
High Involvement 4.21 



No Decision 4.39 Positive Discrepancy 4.39 
Decision 4.34 Negative Discrepancy 4.34 



Note: Analysis of mean responses on combined seven-point scales 

describing the coffeemaker as 1 = "Worse than I expected" to 

7 = "Better than I expected"; expectations about the 

coffeemaker as 1 = "Too high; It was worse than I thought" to 

7 = "Too low; it was better than I thought." Results are for 

a combined scale representing the average response, 

Cronbach's alpha = .57. Significance of results is indicated 

by: a = p< .0001 ; b = p< . 1 ; c = p<.05; d = p< . 1 . 



Table 24 
ANOVA Summary of Coffeemaker Discrepancy 



250 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



1 
1 

I86 



14.83 


1 .07 


0.04 


0.02 


0.11 


0.06 


10.81 


5.48 


0.03 


0.01 


0.16 


0.08 


1.08 


0.55 


2.56 


1 .30 


366.99 





Cell Means (variance) [n] 

1. Low Involvement, No Decision, Positive Discrepancy 4.21 (2.17) [24] 

2. High Involvement, No Decision, Positive Discrepancy 3.76 (2.27) [25] 



3. Low Involvement, No Decision, Negative Discrepancy 4.43 (2.0 

4. High Involvement, No Decision, Negative Discrepancy 3.64 (1.24 

5. Low Involvement, Decision, 

6. High Involvement, Decision, 

7. Low Involvement, Decision, 

8. High Involvement, Decision, 



Positive Discrepancy 4.36 (2.32) 
Positive Discrepancy 3.75 (1.67) 
Negative Discrepancy 4.08 (2.4 



[23] 

:25] 

!25] 
>] 
25] 



Negative Discrepancy 4.04 (1.59) [23] 



Main Effect Means 



Low Involvement 4.27 
High Involvement 3.79 



No Decision 4.00 Positive Discrepancy 
Decision 4.06 Negative Discrepancy 



02 



Note : Based on responses to a scale evaluating expectations 

regarding the coffeemaker as 1 = "Too high; It was worse tl 
I thought" to 7 = "Too low; it was better than I thought." 
Significance of results is indicated by: a = p< . 000 1 ; b = 
p< . 01 ; c = p<.05; 
d = p< . 10 . 



Table 25 
ANOVA Summary of Coffeemaker Expectancy Disconf irmation 



251 



Source 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



df 



Sums of Squares 



185 



5.08 
0.89 

1 .46 
0.35 
0.03 
0.21 
0.73 
1 .34 
289.00 



omega- 
squared 



0.46 
0.57 
0.94 
0.22 
0.02 
0.13 
0.47 
0.86 



Cell Means (variance) [n] 

1. Low Involvement, No Decision, Positive Discrepancy 5.04 

2. High Involvement, No Decision, Positive Discrepancy 4.60 

3. Low Involvement, No Decision, Negative Discrepancy 4.70 

4. High Involvement, No Decision, Negative Discrepancy 4.72 

5. Low Involvement, Decision, Positive Discrepancy 4.60 

6. High Involvement, Decision, Positive Discrepancy 4.74 

7. Low Involvement, Decision, Negative Discrepancy 4.54 
3. High Involvement, Decision, Negative Discrepancy 4.48 

Main Effect Means 



,62) 
,25) 
,13) 
,21) 
■ 42) 
,84) 



[25] 
[25] 
[23] 
[25] 

[25] 
[23] 



39) [24] 
62) [23] 



Low Involvement 4.72 
High Involvement 4.64 



No Decision 4.77 Positive Discrepancy 4.74 
Decision 4.59 Negative Discrepancy 4.61 



Note: Based on responses to a semantic differential scale 

describing the coffeemaker as 1 = "worst than I expected" to 
7 = "Better than I expected". Significance of results is 
indicated by: a = p< . 0001 ; b = p< . 01 ; c = p<.05; d = p< . 1 . 



252 



Table 26 
Selection of Coffee Expectations Scale 



a) Factor Analysis Results 3 

Items Factor 1 Factor 2 

1 Coffee is not bitter .93 

2 Coffee is strong .76 

3 Coffee is quality .68 

4 Coffee is dark . 77 

5 Coffee is good tasting .67 

6 Coffee is expensive .54 



variance explained 39% 23% / 62% 

b) Cronbach's Alpha Computation 

Coffee expectations = mean of (items 1,2,3,4,5,6). 

Pearson Correlation Coefficients: 

Item 2 Item 3 Item 4 Item 5 

.324 

.422 .553 

.398 .487 .466 

.119 .396 .343 .210 

= .73 



Note: a Principal component analysis using varimax rotation 
b Subjects were asked what they expected the coffee to 

be. 
c Using formulation in Carmines and Zeller 1979 







Item 


Item 


2 


-.027 


Item 


3 


.400 


Item 


4 


.255 


Item 


5 


.297 


I tern 


6 


.053 



Table 27 
ANOVA Summary of Coffee Bitterness Expectations 



253 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DF x DI 

Residual 



17.68 
3.13 
0.57 



184 



1 . 
0, 
5, 
6, 
0. 
509. 



,96 
,00 
.64 

38 
. 72 

69 



0.91 

1 .13 
0.21 
0.71 
0.00 
2.04 
2.30 
. 26 



Cell Means (variance) [n] 



Low Involvement, No Dec is 

High Involvement, No Decis 

Low Involvement, ?Jo Decis 

High Involvement, No Decis 

Low Involvement, Decision 

High Involvement, Decision 

Low Involvement, Decision 

High Involvement, Decision 



on, 

on, 

on, 
on, 



Positive Discrepancy 4.54 (3.30) [24] 

Positive Discrepancy 4.44 (2.671 [25] 

Negative Discrepancy 5.27 (2.49; [22; 

Negative Discrepancy 4.24 (2.36) l 2j] 

Positive Discrepancy 4.20 (2.92) [25] 

Positive Discrepancy 4.58 (3.56) [24] 

Negative Discrepancy 4.67 (2.23) [24] 

Negative Discrepancy 4.61 (2.61) [23] 



Main Effect Means 



Low Involvement 4.65 
High Involvement 4.46 



No Decision 4.60 Positive Discrepancy 4.44 
Decision 4.51 Negative Discrepancy 4.68 



Note : Analysis of seven-point scale describing expectations about 
the coffee as 1 = Very Bitter to 7 = Very Not Bitter. 
Significance of results is indicated by: a = p<.0001; b = 
p< . 01 ; c - p<.05; d = p< . 1 . 



254 



quality, it seems reasonable that none would expect bitterness. 
Apparently those subjects in the negative discrepancy (high 
expectations) conditions did not attempt to discriminate degrees of 
"not bitter." 

A composite measure of the remaining coffee expectation scales was 
used as the major indicant of expectations (Cronbach's alpha = .73, 
Table 26). The advertisements and product information given verbally 
to the subjects was essentially unidimensional with respect to product 
attributes. That is, no complex relationships among good and bad 
values were suggested. Thus, the fact that a negatively valenced 
attribute, bitterness, was the only one perceived differently can be 
taken as reassuring evidence that the subjects were attending closely 
to the task. 

Measures of expectations regarding the coffeemaker were generated 
and presented in a similar fashion. All four scales--dif f icult/simple 
to use, fast/slow, difficult/simple to fill, unattractive/attractive-- 
ioaded on one factor in a principal component analysis and were treated 
as measures of one gestalt expectation (Cronbach's alpha = .69, Table 
28). As with the preliminary information regarding the coffee, the 
advertisements and instructions regarding the coffeemaker did not 
provide any discrepant attribute information. 

It should be noted that the type and quantity of information 
provided to the subjects regarding each product were consistent with 
what might occur in the marketplace. The responses to the measures 
discussed above suggest that a single expectation level is generated 



255 



Table 28 
Selection of Coff eemaker Expectations Scale 



a) Factor Analysis Results 3 

Items' 3 Factor 1 

1 Coffeemaker is simple 

to use . 70 

2 Coffeemaker is fast .80 

3 Coffeemaker is simple 

to fill " .77 

4 Coffeemaker is 

attractive .60 



variance explained 52% 

b) Cronbach's Alpha Computation 

Coffeemaker Expectations = mean of (items 1-4) 

Pearson Correlation Coefficients: 

Item 1 Item 2 Item 3 
Item 2 .368 
Item 3 .322 .589 
Item 4 .jo5 .261 .236 

a = .69 
Note: a Unrotated principal component analysis 



256 
from which more specific attribute expectations can be derived. Thus, 
absent any discriminating or discrepant information, the subjects in 
the present study reported highly correlated expectations. The only 
exception to this pattern was a scale that may have been created with 
an inadvertently restricted range. 

Substantive analyses of the composite scales for coffee and 
coffeemaker expectations produced results consistent with the desired 
manipulations. Subjects in negative discrepancy conditions reported 
much higher expectations regarding the coffee, mean score = 5.57, than 
did those in positive discrepancy conditions, mean score = 4.10 
(p< . 0001 , Table 29). This difference explained approximately one- 
third of the overall variance in responses to the composite scale, 
using Hays' omega-squared criterion. An interaction between 
discrepancy manipulation and decision availability accounted for less 
than one percent of the variance and does not qualify the main effect. 

The corresponding analysis of the coffeemaker expectation scale 
was predicted to show no effect of experimental condition. The non- 
significant omnibus F test is adequate evidence of this F 7)1 36 = 1.30, 
p< . 25 , Table 30). As discussed in detail in Chapter 5, only positive 
information was given for the coffeemaker, so the mean rating of its 
expected quality of 5.80 on the 7-point scale lends credence to the 
validity of the procedures and measures. A statistically significant 
difference in mean expectations between the no-decision (5.99) and 
decision (5.61) conditions explained only about 3% of the variance. 
Since this was not predicted and is associated with a non-significant 
omnibus test, it is best considered spurious in this study. 



257 



Table 29 
ANOVA Summary of Composite Coffee Expectations 



Source 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 



df 



Sums of Squares 



F omega- 
squared 



86 



109.32 
104.89 
0.10 
0.02 
3.23 
0.23 
1 .01 
0.07 
203.59 



14.27^ 
95. 83* 
0.09 
0.02 
2.95^ 
0.21 
0.93 
0.07 



331 



.007 



Cell Means (variance) [n. 



No Decis 
No Decis 
No Decis 
No Decis 
Decision 
Decision 
Decision 
Decision 



Positive Discrepancy 4.27 (1.60) [24] 

Positive Discrepancy 4.24 (1.44) [25] 

Negative Discrepancy 5.49 (0.66) [23] 

Negative Discrepancy 5.13 (0.70) [25] 

Positive Discrepancy 3.95 (1.17) [25] 

Positive Discrepancy 4.17 (0.78) [24] 

Negative Discrepancy 5.49 (0.73) [25] 

Negative Discrepancy 5.59 (0.71) [23] 



Main Effect Means 



Low Involvement 4.82 
High Involvement 4.83 



No Decision 
Decision 



4.80 
4.85 



Positive Discrepancy 4.10 
Negative Discrepancy 5.57 



Note: Analysis of mean responses on combined seven-point scales 
describing the expected quality of the coffee as: 
weak/strong in flavor, poor/good quality, light/dark in^ 
color, bad/good tasting, inexpensive/expensive, and assigned 
values of one/seven respectively. Results are for a combined 
scale representing mean response across the scales, 
Cronbach's alpha = .73. Significance of results is indicated 
by: a = p<.0001; b = p< . 1 ; c = p<.05; d = p<.10. 



253 



Table 30 
ANOVA Summary of Coffeemaker Expectations 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 
Decision (DE) 
Involvement (I) 
DE x DI 
I x DI 
I x DE 
I x DE x DI 
Residual 186 



Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 



9.91 
0.85 

7.23 
0.40 
0.17 
0.01 
0.60 
0.79 



1.30 
0.78 
6.65* 

0.37 
0.16 
0.01 
0.56 
0.72 



,03 



202.16 



Cell Means (variance) [n] 

No Decision, Positive Discrepancy 5.99 (0.97) [24] 

No Decision, Positive Discrepancy 5.93 (1.23) [25] 

No Decision, Negative Discrepancy 6.21 (0.49) [23] 

No Decision, Negative Discrepancy 5.86 (0.98) [25] 

Decision, Positive Discrepancy 5.56 (0.90) [25] 

Decision, Positive Discrepancy 5.47 (1.99) [24] 

Decision, Negative Discrepancy 5.64 (1.22) [25] 

Decision, Negative Discrepancy 5.77 (0.87) [23] 



Main Effect Means 



Low Involvement 5.84 
High Involvement 5.76 



No Decision 
Decision 



5.99 
5.61 



Positive Discrepancy 5.74 
Negative Discrepancy 5.86 



Note : Analysis of mean responses on combined seven-point scales 
describing the expected quality of the coffeemaker as very 
difficult/very simple to use, very slow/very fast, very 
difficult/very simple to fill and unattractive/attractive. 
Results are reported for a combined scale representing the 
mean of the four items, Cronbach's alpha = .69. Significance 
of results is indicated by: a = p<.0001; b = p<.01; c = 
p<.05; 
d = p<. 10. 



259 
The apparent weakness in the discrepancy manipulation rests in the 
subjects' perception of product quality (possibly due to individual 
differences in taste discrimination — Buchanan, Givon and Goldman 1987). 
Measures of perception did not reveal the simple main effect that was 
predicted based on objective quality evaluations of the target 
products. Detailed analyses of product perception results suggested 
that individual differences in flavor perception sensitivity were 
responsible for the deviation from expectations. 4 

The questionnaire was constructed with the assumption that the 
five coffee perception scales could be averaged to serve as the sole 
measure of coffee perception. 5 Similarly, the coffeemaker perception 
was expected to be unidimensional ; enabling the use of means across 
four scales. Principal component factor analyses, using a varimax 
rotation, did not return these patterns. Rather, the coffee scales 
seemed to describe three factors that were labeled as follows for 
convenience: Flavor (mean of bitter/not bitter and strong/weak in 
flavor scales); Quality (mean of poor/good quality and bad/good tasting 
scales); and Color (light/dark in color scale) (Table 31). The 
coffeemaker perceptions also appeared to have been the result of 
separable evaluations: Speed (slow/fast scale); Fill (difficult/simple 



4 Analyses of satisfaction and attitude scales, reported below, 
do show a similarly weak but significant effect of the discrepancy 
manipulation. These provide some assurance that the desired effect is 
present, albeit either not as strong or as sensitively measured as 
would have been preferred. 

5 No companion measure to the inexpensive/expensive expectation 
scale was attempted since no additional information was made available 
to the subjects regarding this attribute. The "worse/better than I 
expected" item which appeared in the same section as the perception 
scales, Appendix C, was discussed above. 



260 



Table 31 
Selection of Coffee Perception Scales 



a) Factor Analysis Results 3 

Items Factor 1 Factor 2 

1 Coffee is not bitter .59 

2 Coffee has weak flavor -87 

3 Coffee is good quality .78 

4 Coffee is dark color ■ -.32 

5 Coffee is good tasting .81 



6 Coffee is better than b 
expected .86 

7 Expectations too low b .78 



variance explained 41% 1 9%/ 60% 



b) Cronbach's Alpha Computation 

Flavor = mean of (item 1, item 2) 
Quality = mean of (item 3, item 5) 
Color d = item 4 

Pearson Correlation Coefficients: 

Item 1 , item 2 = .322; a = .49 
Item 3, item 5 = .500; a = .67 



Note: a Principal component analysis using varimax rotation 

b These items weren't designed to be part of the perception 

scale, included as a check of confounding between perception 

and disconf irmation. 
c Using formulation in Carmines and Zeller 1979. 
d This item correlated too weakly to be included in either 

scale. 



261 



to fill scale); Simple (difficult/simple to use scale); and 
Attractiveness (unattractive/attractive scale) (Table 32). 

Among the three coffee perception scales, two did not appear 
responsive to experimental manipulations. Neither Quality (F 7)188 = 
1.55) nor Color (F 7jl84 = 0.93) demonstrated effects that were 
significantly explained by the ANOVA model (Tables 33 and 34, 
respectively) . Effects within each model (a response involvement by 
discrepancy interaction for coffee quality and a response involvement 
main effect for coffee color) did reach statistical significance, but 
were relatively weak and should not be given a great deal of credence 
in light of non-significant omnibus F-tests. 6 

The Flavor scale did exhibit significant effects of the 
manipulations. Specifically, the coffee in the positive discrepancy 
conditions was perceived to be somewhat weaker than that in the 
negative discrepancy conditions. This main effect is qualified by a 
significant interaction between decision making and discrepancy. When 
the subject did not decide on the coffee to sample, the difference in 
evaluation of flavor strength (M = 4.13 for positive discrepancy, M = 
3.95 for negative discrepancy) is not significant. Having decided on 
the brand of coffee to sample, subjects rated the coffee in the 



5 It is interesting to note the apparent effects for future 
investigation. The interaction seemed to indicate differential 
discrimination in perception depending on focus of involvement. Under 
"high" involvement, the coffee was perceived to be of higher quality ir 
the positive discrepancy conditions, M = 5.3 than in the negative 
discrepancy conditions, M = 4.6. On the other hand, under "low" 
involvement subjects in the positive discrepancy conditions perceived 
the coffee to be of no better quality M = 4.92 than did subjects in the 
negative discrepancy conditions M = 5.24, the latter means are not 
significantly different at a = .05 using Tukey's Honestly Significant 
Difference test. 



262 



Table 32 
Correlations Among Coffeemaker Perception Scales 



Items 

1 Simple to use 

2 Fast 

3 Simple to fill 

4 Attractive 

Pearson Correlation coefficients: 

Item 1 Item 2 Item 3 
Item 2 .053 
Item 3 .278 .358 
Item 4 .163 .039 .073 



Note: Each scale was analyzed and is reported separately due to low 
correlations . 



Table 33 
ANOVA Summary of Coffee Quality Perception 



263 



Source 



df 



Sums of Squares 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



Cell Means (variance) [n. 



18 


92 


1 


80 





28 





96 


2 


60 


12 


91 





16 





00 


328 


30 



1 .55 
1 .03 
0. 16 
0.55 
1 .49 
7.39'c 
0.09 
0.00 



omega- 
squared 



Low Involvement, No Decision, Positive Discrepancy 5.10 (1 

High Involvement, No Decision, Positive Discrepancy 5.42 (1 

Low Involvement, No Decision, Negative Discrepancy 5.20 (1 

No Decision, Negative Discrepancy 4.48 (1 

Decision, Positive Discrepancy 4.74 (1 

Decision, Positive Discrepancy 5.17 (1 

Decision, Negative Discrepancy 5.29 (2 



High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 



High Involvement, Decision, Negative Discrepancy 4.70 (2 



Main Effect Means 



,03 



79) [25] 

26) [25] 

52) [251 

7c) [25] 

92) [25] 

20) [241 

06) [26] 

36) [23] 



Low Involvement 5.08 
High Involvement 4.94 



No Decision 5.05 Positive Discrepancy 5.11 
Decision 4.98 Negative Discrepancy 4.92 



Note: Analysis of mean responses on combined seven-point scales 
describing the perceived quality of the coffee as very 
poor/very good quality and very bad/very good tasting, 
assigned values of one and seven, respectively. Results are 
for a combined scale representing mean response to two items, 
Cronbach's alpha = .67. Significance of results is indicated 
by: a = p<.0001; b = p< . 01 ; c = p<.05; d = p< . 1 . 



Table 34 
ANOVA Summary of Coffee Color Perception 



264 



Source 



df 



Sums of Squares 



omega- 
jquared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



184 



1 .87 


0.93 


0.83 


0.41 


0.62 


0.31 


8.75 


4.37 = 


2.69 


1 .34 


0.16 


0.08 


0.04 


0.02 


0.06 


0.03 


368.59 





.02 



Cell Means (variance) [n 



Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 



No Decis 
No Decis 
No Decis 
No Decis 
Decision 
Decision 
Decis ion 
Decision 



ion, Positive Discrepancy 5.42 (2.08) [24] 

on, Positive Discrepancy 5.72 (1.38) [25] 

on, Negative Discrepancy 5.22 (3.45) [23] 

on, Negative Discrepancy 5.71 (1.52) [24] 

Positive Discrepancy 5.00 (2.00) [24] 

Positive Discrepancy 5.43 (1.62) [23] 

Negative Discrepancy 5.35 (2.48) [26] 

Negative Discrepancy 5.83 (1.51) [23] 



Main Effect Means 



Low Involvement 5.25 
High Involvement 5.67 



No Decision 5.52 Positive Discrepancy 5.40 
Decision 5.40 Negative Discrepancy 5.52 



Note : Analysis of responses to a seven-point scale describing the 
prepared coffee as 1 = "Very light in color" to 7 = "Very 
dark in color". Significance of results is indicated by: a 
= p< . 0001 ; b = p< . 01 ; c = p<-05; d = p< . 1 . 



265 

positive discrepancy conditions as weaker in flavor, M = 4.32, than 
did subjects rating the negative discrepancy product, M = 3.38 (using 
Tukey's Honestly Significant Difference test a = .05). Note, however, 
that this interaction accounts for only about 2% of the variance in the 
results (Table 35). 

Because the correlation of the two measures used for the Flavor 
scale was so low (Cronbach's a = .49), the individual scales were also 
examined separately. Alone, the bitterness scale did not respond to 
the manipulations (F 7)186 = 1.69, p< . 1 1 , Table 36). The apparent 
perception of the peanut blend as more bitter, mean = 4.29, than the 
packaged coffee, mean = 4.86 was appropriate — even if the omnibus test 
did not reach a conventional significance level. 

The perception of flavor strength was more complex; the three-way 
interaction among the manipulations is depicted in Figure 12. The 
interaction explains only about one percent of the variance, but 
qualifies the interaction between decision and discrepancy as well as 
the discrepancy main effect (Table 37). For the positive discrepancy 
conditions, only the low involvement, no decision group perceived the 
coffee to be stronger, mean - 2.96, than did the other subjects (based 
on Tukey's Honestly Significant Difference text, a = .05). In the 
negative discrepancy conditions, the low involvement, no decision 
group was again the outlier. In this case, however, the blend was 
rated less strong, mean = 3.64, by those in the low involvement, no 
decision condition (the high involvement, no decision mean, 3.2, was 
not significantly different from either 3.64 or its high involv.ement, 
decision counterpart, 2.78). 



Coffee 
strength 

Very 7.0 r 
Weak 



266 



6.0 



5.0 



4.0 



3.0 



Very 2.0 
Strong 



1.0 



0.0 



Low 
involvement 



High 

involvement 



J L 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 

Figure 12 

Perceptions of Coffee Strength 

Interaction Among Experimental Conditions 



Table 35 
ANOVA Summary of Coffee Flavor Perception 



257 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 
Decision (DE) 



Involvement 
DE x DI 
I x DI 
I x DE 
I x DE x 
Residual 



(I) 



DI 



32.56 


2.75^ 


15.01 


8.88^ 


1 .96 


1 .16 


0.33 


0.20 


7.26 


4.29= 


3.49 


2.07 


0.01 


0.00 


4.53 


2.68 


317.82 





.02 



Cell Means (variance) [n] 



Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 



No Dec is 
No Decis 
No Decis 
No Decis 
Decision 
Decision 
Decision 
Decision 



on, Positive Discrepancy 3.88 (1.21) [25] 

on, Positive Discrepancy 4.33 (2.11) [25] 

on, Negative Discrepancy 4.28 (1.77) [23] 

on, Negative Discrepancy 3.64 (1.09) [25] 

Positive Discrepancy 4.38 (2.19) [25] 

Positive Discrepancy 4.25 (1.48) [24] 

Negative Discrepancy 3.40 (1.28) [26] 

Negative Discrepancy 3.35 (2.46) [23] 



Main Effect Means 



Low Involvement 
High Involvement 



3. 



No Decision 
Decision 



4.04 Positive Discrepancy 
3.85 Negative Discrepancy 3.6 



22 



Note : Analysis of mean responses to two seven-point scales 

describing the prepared coffee as 1 = "Very Strong in Flavor" 
to 7 = "Very Weak in Flavor" and 1 = "Very Bitter" to 7 = 
"Very Not Bitter", Cronbach's alpha = .49. Significance of 
results is indicated by: a = p<.0001; b = p< . 01 ; c = p<.05; d 
= p<. 10. 



Table 36 
ANOVA Summary of Coffee Bitterness Scale 



2b8 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



7 
1 
1 
1 
1 
1 
1 
1 
18b 



34.27 
15.13 



49 

,02 

33 



4.92 

0.47 

3.33 

539.07 



1 .69 
5.22= 
1 .20 
1 .73 
0.82 
1 .70 
0.16 
1.15 



,02 



Cell Means (variance) [n] 

Low Involvement, No Decision, Positive Discrepancy 4.80 (3.33) [25] 

High Involvement, No Decision, Positive Discrepancy 4.96 (3.12) [25] 

Low Involvement, No Decision, Negative Discrepancy 5.04 (2.86) [23] 

High Involvement, No Decision, Negative Discrepancy 4.04 (2.48) [24] 

Low Involvement, Decision, Positive Discrepancy 4.92 (2.95) [24] 

High Involvement, Decision, Positive Discrepancy 4.75 (2.72) [24] 

Low Involvement, Decision, Negative Discrepancy 4.19 (2.64) [26] 

High Involvement, Decision, Negative Discrepancy 3.91 (3.08) [23] 



Lew Involvement 4.72 
High Involvement 4.43 



Main Effect Means 

No Decision 4.71 Positive Discrepancy 4.86 
Decision 4.44 Negative Discrepancy 4.29 



Note : Analysis of a scale describing the coffee as 1 = Very Bitter 
to 7 = Very Not Bitter. Significance of results is indicated 
by: a = p<.0001; b = p< . 01 ; c = p<.05; d = p< . 1 . 



Table 37 
ANOVA Summary of Coffee Flavor (single) Scale 



269 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



187 



44.54 

14.65 

0.85 

0.49 

16.76 

2.68 

0.50 

7.91 

426.46 



Cell Means (variance) [n] 



2.79b 
6.42 = 
0.37 
0.21 
7.35^ 
1 . 18 
0.22 
3.47^ 



.026 



.031 



.012 



Low Involvement, No Decision, Positive Discrepancy 2.96 (2.21) [25] 

High Involvement, No Decision, Positive Discrepancy 3.80 (3.33) [25] 

Low Involvement, No Decision, Negative Discrepancy 3.64 (1.86) [22] 

High Involvement, No Decision, Negative Discrepancy 3.20 (1.67) [25] 

Low Involvement, Decision, Positive Discrepancy 3.92 (2.74) [25] 

High Involvement, Decision, Positive Discrepancy 3.75 (2.11) [24] 

Low Involvement, Decision, Negative Discrepancy 2.62 (1.45) [26] 

High Involvement, Decision, Negative Discrepancy 2.78 (2.91) [23] 

Main Effect Means 



Low Involvement 3.27 
High Involvement 3.39 



No Decision 
Decision 



3.39 Positive Discrepancy 3.61 
3.27 Negative Discrepancy 3.04 



Note : Analysis of a scale describing the coffee as 1 = Very Strong 
in Flavor to 7 - Very Weak in Flavor. Significance of 
results is indicated by: a = p< . 0001 ; b = p< . 01 ; c = p<.05; d 
= p<. 10. 



270 

It is interesting that the group that was assigned a "non- 
critical" test product perceived it differently in each case. This 
could be due to a number of processes: a stronger assimilation effect; 
greater attribution to the coffeemaker, and the like. While it is 
tempting to speculate on such causes, the fact that the pattern 
emerges for one measure and with relatively small effect sizes elicits 
caution. It is sufficient to note that the general pattern of flavor 
strength ratings was consistent with the product formulation 
manipulation. 

Among the coffeemaker perception scales (no correlation was high 
enough to justify combining items, Table 32), only the one showing 
evaluations of speed revealed significant effects explainable by the 
experimental manipulations (Tables 38, 39, 40, 41). Although the 
variance explained was small, each manipulation demonstrated a 
significant main effect: discrepancy, 2% of variance, decision, 2% of 
variance, and involvement 1% of variance. Speed was perceived to be 
higher (coffeemaker was faster) in the positive discrepancy, M = 5.37, 
no decision, M = 5.36, and high response involvement (with coffee), M = 
5.31 conditions than in the negative discrepancy, M = 4.75, decision 
present, M = 4.78, and low involvement, M = 4.83, conditions. No 
manipulation effects were hypothesized, but speculative interpretations 
of these data may be borne out in future studies. The involvement 
effect might signal some effectiveness of that manipulation — subjects 
were more critical when the coffeemaker was the target product. The 
difference in beverage formulation may have affected the actual speed 
and therefore perception of speed between discrepancy conditions-- 



Table 38 
ANOVA Summary of Coffeemaker Speed Perception 



271 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



49, 
18. 
15, 



10.47 
0.37 
2.91 
0.75 
0.00 
>32 . 35 



2.11<= 
0.02 c 
67 = 
11 d 
11 
87 
22 



0.00 



,02 
.02 
,01 



Cell Means (variance) [n] 

Low Involvement, No Decision, Positive Discrepancy 5.28 (4.13) [25] 

High Involvement, No Decision, Positive Discrepancy 6.12 (1.36) [25] 

Low Involvement, No Decision, Negative Discrepancy 4.83 (4.51) [23] 

High Involvement, No Decision, Negative Discrepancy 5.16 (3.47) [25] 

Low Involvement, Decision, Positive Discrepancy 4.76 (3.52) [25] 

High Involvement, Decision, Positive Discrepancy 5.33 (2.41) [24] 

Low Involvement, Decision, Negative Discrepancy 4.46 (3.46) [26] 

High Involvement, Decision, Negative Discrepancy 4.57 (4.17) [23] 

Main Effect Means 



Low Involvement 4.83 
High Involvement 5.31 



No Decision 5.36 Positive Discrepancy 5.37 
Decision 4.78 Negative Discrepancy 4.75 



Note : Analysis of responses to a seven-point scale describing the 
coffeemaker as 1 = "Very Slow" to 7 = "Very Fast". 
Significance of results is indicated by: a = p< . 000 1 ; b = 
p< . 01 ; c = p<.05; d = p< . 10. 



Table 39 
ANOVA Summary of Coffeemaker Filling Perception 



272 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



,89 
,58 
,11 
,65 
,73 
09 
,13 



187 



0.98 
542.03 



0.57 
1 .68 
0.25 
0.14 

0.83 
0.91 
0.03 
0.22 



Cell Means (variance) [n] 

1. Low Involvement, No Decision, Positive Discrepancy 4.76 (5.94) [25] 

2. High Involvement, No Decision, Positive Discrepancy 5.36 (5.41) [25] 

3. Low Involvement, No Decision, Negative Discrepancy 5.86 (4.31) [22] 

4. High Involvement, No Decision, Negative Discrepancy 5.60 (3.42) [25] 

5. Low Involvement, Decision, Positive Discrepancy 5.08 (4.91) [25] 

6. High Involvement, Decision, Positive Discrepancy 5.29 (4.22) [24] 

7. Low Involvement, Decision, Negative Discrepancy 5.35 (3.36) [26] 

8. High Involvement, Decision, Negative Discrepancy 5.26 (4.47) [23] 

Main Effect Means 



Low Involvement 5.24 
High Involvement 5.38 



No Decision 5.38 Positive Discrepancy 5.12 
Decision 5.24 Negative Discrepancy 5.51 



Note: Analysis of responses to a seven-point scale describing the 
coffeemaker as 1 = "Very Difficult to Fill" to 7 = "Very 
Simple to Fill". Significance of results is indicated by: i 
= p<.0001; b = p< . 01 ; c = p<.05; d = p< . 1 . 



273 



Table 40 
ANOVA Summary of Coffeemaker Use Perception 



Source df Sums of Squares F omega- 

squared 



Model 7 8.90 0.65 

Discrepancy (DI) 1 0.21 0.11 

Decision (DE) 1 3.80 1 .94 

Involvement (I) 1 0.02 0.01 

DE x DI 1 1.93 0.99 

I x DI 1 0.02 0.01 

I x DE 1 0.06 0.03 

I x DE x DI 1 2.91 1 .49 

Residual 188 367.77 

Cell Means (variance) [n] 



Low Involvement, No Decision, Positive Discrepancy 6.12 (2.44) [25 

High Involvement, No Decision, Positive Discrepancy 6.36 (1.41) [25. 

Low Involvement, No Decision, Negative Discrepancy 6.61 (1.61) [23 

High Involvement, No Decision, Negative Discrepancy 6.40 (1.50) [25. 

Low Involvement, Decision, Positive Discrepancy 6.32 (0.98) [25. 

High Involvement, Decision, Positive Discrepancy 6.00 (3.39) [24 

Low Involvement, Decision, Negative Discrepancy 5.92 (2.23) [26 

High Involvement, Decision, Negative Discrepancy 6.13 (2.12) [23 

Main Effect Means 



Low Involvement 6.23 No Decision 6.36 Positive Discrepancy 6.20 
High Involvement 6.23 Decision 6.09 Negative Discrepancy 6.26 



Note : Analysis of responses to a seven-point scale describing the 
coffeemaker as 1 = "Very Difficult to Use" to 7 = "Very 
Simple to Use". Significance of results is indicated by: a 
p<.0001; b = p< .01 ; c = p<.05; d = p< . 1 . 



Table 41 
ANOVA Summary of Coffeemaker Appearance Perception 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



187 



21.37 


1 .32 


0.64 


0.28 


1 .07 


0.46 


3.00 


1 .30 


6.98 


3.02d 


3.78 


1 .64 


•2.11 


0.92 


3.42 


1 .48 


431 .58 





Cell Means (variance) [n] 

Low Involvement, No Decision, Positive Discrepancy 5.72 (1.29) [25] 

High Involvement, No Decision, Positive Discrepancy 4.72 (2.54) [25] 

Low Involvement, No Decision, Negative Discrepancy 4.91 (2.81) [23] 

High Involvement, No Decision, Negative Discrepancy 5.00 (3.17) [25] 

Low Involvement, Decision, Positive Discrepancy 4.72 (2.96) [25] 

High Involvement, Decision, Positive Discrepancy 4.67 (2.41) [24] 

Low Involvement, Decision, Negative Discrepancy 5.20 (1.50) [25] 

High Involvement, Decision, Negative Discrepancy 5.17 (1.79) [23] 

Main Effect Means 



Low Involvement 5.14 
High Involvement 4.89 



No Decision 5.09 Positive Discrepancy 4.96 
Decision 4.94 Negative Discrepancy 5.07 



Note : Analysis of responses to a seven-point scale describing the 
coffeemaker as 1 = "Very Unattractive" to 7 = "Very 
Attractive". Significance of results is indicated by: a = 
p< . 0001 ; b = p< . 01 ; c = p<.05; d = p< . 1 . 



275 
although this was not noticeable to the experimenter. 7 Finally, having 
made a decision to try a particular product, subjects may have been 
more anxious to assess their decision — and therefore less patient with 
the coffeemaker — than those who did not choose the beverage product. 
Other post hoc interpretations are possible; these account for the 
results (which may be found to be spurious on replication). 

The data relating to the discrepancy manipulation reveal more 
complexity than originally intended. The advertisements and 
instructions communicated differential expectations quite well. One 
trial of the actual coffee beverage appeared to have been less 
convincing in the face of the prior testimony. The ratings that 
resulted, however, were in appropriate directions and generally 
significant (albeit weakly so). As appropriate, there were few such 
differences for analogous measures dealing with the "decoy" product, 
the coffeemaker. In both cases, however, the subjects appeared to have 
carefully weighed the information newly acquired in product trial 
against that which was presented "and the greater dimensionality" to 
them. This was evidenced by the difference in discrimination among 
individual scales. 

Involvement . Since the response involvement manipulation was 
based on the one used by Petty et al. (1983), it would seem natural to 
replicate the manipulation check measure as well. However, "asking 
subjects what gift they had been told to expect" (p. 140) reduces to a 



7 Also, it should be recalled that identical — but different — 
coffeemakers were used to avoid contamination. There may have been 
slight differences in operating characteristics between discrepancy- 
conditions . 



276 
extraneous measure when — as is the case in the present study — the 
subjects actually have a gift in hand and are unaware that they may be 
offered the opportunity to exchange it after completing the 
questionnaire. Thus a position could be taken that response 
involvement was successfully manipulated whenever procedures were 
followed properly. This is similar to the argument raised for the 
decision manipulation and fully consistent with the implementation of 
the construct by Petty et al. 8 

In an effort to avoid adopting such a dogmatic stance, 
manipulation check measures adapted from Sherrell and Shimp (1982) were 
used (Appendix C) . At best, this selection was ill-advised. However, 
these were the only measures located that claimed any relevance to 
response involvement (note that the Petty et al. measure does not 
directly tap the construct) . Analyses of the scales which rated 
"evaluation of the cof f ee/cof f eemaker" as: interesting/not interesting 
to me, important/not important to me, difficult/not difficult to 
evaluate, took/did not take a lot of thought, task was/was not very 
involving, and I put a lot/no thought into evaluation, were conducted 
on individual as well as aggregate scales. Once again, the expected 
unidimensionality did not obtain (Tables 42, 43) and sub-scales were 
created. 



8 As originally designed, the procedure for the present study 
mirrored the Petty et al. formulation more closely. During the 
pretests, it became evident that it was necessary to award the "gift" 
prior to product trial in order to convince subjects of the veracity of 
the offer. This alteration in the protocol rendered the Petty et al. 
check measure redundant at best. No analogous measure could be derived 
that would not arouse subject suspicion. 



Table 42 
Selection of Coffee Involvement Scale Combinations 



277 



a) Factor Analysis Results 3 





Items b 


1 


Interesting 


2 


Important 


3 


Difficult 


4 


Took thought 


5 


Involving 


6 


Put in thought 



Factor 1 
.78 
.76 



.46 



Factor 2 



63 
74 
76 



variance explained 



28% 



27% / 55% 



b) Cronbach's Alpha Computation 

Pearson Correlation Coefficients: 

Item 2 Item 3 



Item 2 
Item 3 
Item 4 
Item 5 
Item 6 



Item 1 
.439 

-. 147 
.138 
.005 
.151 



-.127 
.237 
. 185 
.156 



.249 

.227 

-.134 



Item 4 



+ 01 



.1. 



Combined scales 

mean of item 1, item 2 a = .61 

mean of item 3, item 4, item 5 a = .55 



Item 5 



.196 



Note : a Principal component analysis using varimax rotation 
b Subjects were asked to rate the evaluation of the 

coffee. 
c Using formulation in Carmines and Zeller 1979. 



27i 



Table 43 
Selection of Coffeemaker Involvement Scale Combinations 



a) Factor Analysis Results 3 

Factor 1 Factor 2 
.84 
.74 
-.65 

.82 
.81 
.46 





Items b 


1 


Interesting 


2 


Important 


3 


Difficult 


4 


Took thought 


5 


Involving 


6 


Put in thought 



variance explained 32% 28% / 60% 

b) Cronbach's Alpha Computation 

Pearson Correlation Coefficients: 

Item 3 Item 4 Item 5 



.161 

.178 .449 

.230 .199 .286 





Item 1 


Item 


Item 2 


.594 




Item 3 


-.312 


-.203 


Item 4 


.025 


.207 


Item 5 


.158 


-.008 


Item 6 


.253 


.218 



Combined scales 

mean of item 1, item 2 a = .75 

mean of item 4, item 5 a = . 62 



Note: a Principal component analysis using vanmax rotation 
b Subjects were asked to rate the evaluation of the 

coffeemaker. 
c Using formulation in Carmines and Zeller 1979. 



279 



Analyses of variance performed on the scales, both individually 
and in sets, yielded few significant results detectable by omnibus F 
statistics (most F's < 1, Table 44). The lack of any discernible 
patterns among or within the models that did reach a conventional 
level of significance suggests that these may have been due to chance 
and might therefore be considered spurious. 

At this point, it is appropriate to consider deeming the measures 
inadequate rather than to conclude that the manipulation was non- 
existent, albeit recognizing that this creates a tenuous philosophical 
position. However, such a conclusion can be defended on several 
grounds. Most to the point is the fact that the scale's developers had 
similar difficulty detecting a within product difference, "manipulation 
did generate different degrees of cognitive activity in the high- and 
low-involvement groups, but that the two verbal methods were simply 
incapable of detecting this" (Sherrell and. Shimp 1982, p. 107). The 
authors further opined that the lack of sensitivity of the verbal 
measures may have been due to decay during the interval between 
manipulation and measurement. A chronometric measure (decision time) 
did yield the results expected for the manipulation, providing some 
confirmation of this conjecture. 

Others have despaired the lack of adequate measures, "the 
reactive nature of research procedures can confound the measurement of 
response involvement. More [sic] participation in research can 
heighten the involvement level of a subject; artificially high levels 
of response involvement may result" (Houston and Rothschild 1978b, p. 
186). While heightened involvement did not seem to be a particular 



o u — > 

- o> 

c <r oo 

« OH 



o 


so 


0> 




CN 


CO 


c 


o 


i-4 


03 






cu 






X 


u 





vO 00 

14-1 4_> O rM 

3 ^ 



0) 




■H n co 

cm U .— 1 


,0 

03 
H 


01 
CO 


CM w 
•H 

Q 



ooooo^oo 



cn o O m to O h 



o o o © o 



OOOCNO^OO 



OCNO'-iOOOO 



HC0^OHO\C^C0N 
f-HOO'— I O <J <t H f) 

cn cm — i o n * N N 



HC0MHO>HHOV 

mHi/i^OM<fH 





_l 


i-i w 


4) 


o 


o n 




J2 •• 




03 
U 


l? 




CO 


CO ^-' 







CU 


•w 


> 


a 


HH 


W 


W 


^-j 


w 


en 


■ — i 




Q 


Q 


Q 


0) 


o 


•H 





X 








■o 


to 


u 


> 




X 


X 


X 





■M 


u 


G 


w 








X 


a 


Q 


M 


a 


M 


H 


1-1 



aj 


2 


^ 


U 






X 


2 




CM 




^-v 





ao 


a> 




CN 


CO 


C 


z 


rH 



U N 00 

u X ~> 
o -^ 



a h » 

0) Z i-H 



OOhqhcnOO 



O—iOOOO— <o 



hi— i o O O cn m o 



OHOOO— 'OO 



OCNOOOcNOO 



oooooooo 



i— i o O h o <n o 



■• — 1 


M W ■—> 


O 


a a m 


^* _a .. 




co E ^ 




o >, c 




CO CO w 





p-4 M 03 



-a so o > xxx 

o -h cu c w 

ZQQchQmwi-i 



280 



to ca 
co .3 j: 
cu a. a. 



.— < co 



—i x J= 



JJ 


oi 


u 




— i 


m 


,-H 


>-. 


CO 




sO 


r«. 


a 


-C 


■H 








en 


a 


— 








cu 




nj 


at 


II 


il 


u 


u 


i-j 


c 








o 


CO 


•H 


03 


03 


U-l 


cm 




■v> 


j3 


— 


o 




b 


CO 


0. 


a 




cn 




OJ 






4) 




OJ 


X 


03 


03 


CJ 




M 








3 


•H 


03 


* 


CO 


CO 


ta 


-3 




u 






U 




-o 


03 


-3 


jC 


•-4 


x: 


CU 


CU 


O 


U 


CM 


X 


M 




03 


03 


•M 


w 


H 


u 


J 


J2 


C 







cu 


3 


3 


o£ 


0) 


a 


-o 








•M 


CU 


0) 


c 


- 


H 


CO 


CO 


OtJ 


3 


U 


CJ 



41 CM 00 'H 



281 
problem in the present study (means were generally around the middle 
range — 3 to 5 — of the 7-point scales), a restriction of range due to a 
ceiling effect would be one explanation for the lack of variance among 
conditions . 

In summary, the measures of the response involvement measure 
appear to have been flawed. The manipulation itself, however, is well- 
grounded in both prior research and self-reports of subjects in the 
pretests reported herein. This can be taken as presumptive evidence 
that the manipulation was executed as defined. This suggests that a 
conservative approach to data analysis is appropriate, without losing 
completely the response involvement factor. As argued by Sternthal, 
Tybout and Calder, "manipulation checks deserve no special status in 
rigorous tests of theory (1987, p. 119). That is, negative or 
ambiguous responses on these scales should not be weighted more heavily 
than responses on any other scales. This issue is discussed further in 
Chapter 7. A conservative approach to the data is in order. The 
response involvement factor and its interactions remain in the 
analyses reported below. Discussion of its effects — especially where 
specific hypotheses were made--is constrained in recognition of the 
problem with demonstrating conclusively the presence of a strong 
manipulation. 
Tests of Hypotheses 

The manipulations defined an experiment that was designed to 
evaluate the utility of the extended conceptualization via application 
of the measures of emotions. That is, the satisfaction measures 
derived from the DES should show effects attributable to the 



282 



manipulated additional constructs, decision and response involvement. 
Other measures — perception, attitude, common satisfaction — should be 
consistent with other studies conducted under the disconf irmation 
paradigm. 

Attitude measures . Among the most basic responses to a product 
evaluation is thought to be the development or modification of an 
attitude. Hypothesis 3 asserted that attitude measures would be 
insensitive to the decision opportunity and response involvement 
manipulations. The pattern of means anticipated is depicted in Figure 
11b. 

As noted in the previous chapter, a six-item scale was devised to 
assess attitude toward the coffee. A principal components factor 
analysis demonstrated only the one dimension (obviating the need for 
rotation) explaining 87% of variance, which enabled the use of the mean 
across the scales as an attitude index (Cronbach's a = .97, Table 45). 

In this case, the model did explain a significant, albeit not 
large, proportion of the variance in the criterion (F 7;181 = 1.89, p < 
.10, Table 46). The majority of this explanatory power lies in the 
discrepancy manipulation (F 1>181 = 7.96, p < .01, omega-squared = 
.04). Subjects in positive discrepancy conditions had a more favorable 
attitude, M = 5.30, than did subjects in negative discrepancy 
conditions, M = 4.67. This finding is particularly interesting given 
the somewhat discouraging results revealed by the discrepancy 
manipulation checks. 



283 



Table 45 
Reliability of Coffee Attitude Scale 



a) Factor Analysis Results 3 

Factor 1 
.94 
.96 
.91 
.85 
.95 
.97 





Items 


1 


Good 


2 


Satisfactory 


3 


Favorable 


4 


Valuable 


5 


Like 


6 


Approve 



variance explained 



575 



b) Cronbach's Alpha Computation 13 

Pearson Correlation Coefficients: 





Item 1 


Item 


Item 2 


.891 




Item 3 


.801 


.889 


Item 4 


.769 


.745 


Item 5 


.900 


.894 


Item 6 


.892 


.928 



Item 3 



.704 
.825 
.857 



Item 4 Item 5 



.763 

.806 .904 



97 



Note: a Principal component analysis. 

b Using formulation in Carmines and Zeller 1979. 



Table 46 
ANOVA Summary of Coffee Attitude Ratings 



284 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



181 



32.18 
19.37 
1 .75 
0.59 
8.71 
0.83 
0.02 
0.89 
440.29 



Cell Means (variance) [n_ 



1 .89^ 

7.96b 

0.72 

0.24 

3.58^ 

0.34 

0.01 

0.37 



Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 



Positive Discrepancy 5.55 (2.51) 

Positive Discrepancy 5.68 (1.25) 

Negative Discrepancy 4.47 (3.04) 

Negative Discrepancy 4.61 (2.31) 

Positive Discrepancy 4.81 (2.34) 

Positive Discrepancy 5.17 (2.48) 

Negative Discrepancy 4.87 (2.35) 

Negative Discrepancy 4.69 (3.47) 



Main Effect Means 



04 



01 



[25] 
[25] 
[17] 
[25] 
[25] 
[24] 
[25] 
[23] 



Low Involvement 4.96 
High Involvement 5.04 



No Decision 5.13 Positive Discrepancy 5.30 
Decision 4.88 Negative Discrepancy 4.67 



Note : Analysis of mean responses to combined seven-point semantic 
differential scales describing the respondent's feelings 
about the coffee as bad/good, unsatisfactory/satisfactory, 
unfavorable/favorable, worthless/valuable, dislike/like, 
disapprove/approve; scored one/seven respectively. Results 
are for a combined scale representing the average response, 
Cronbach's alpha = .97. Significance of results is indicated 
by: a = p< . 0001 ; b = p< . 1 ; c = p<.05; d = p< . 1 . 



285 
It may be that the subjects couldn't express, or weren't given an 
opportunity to express, themselves prior to this stage of the 
questionnaire. That is, it was believed that the perception and 
disconf irmation scales measured important, salient attributes. Either 
subjects could not identify the specific cause of displeasure or they 
were not provided with a measure appropriate to their need. Thus, the 
manipulation check measures may, again, have been weaker than the 
manipulation. Also, it should be noted that even in the negative 
discrepancy conditions, the mean attitude was slightly favorable. 
Further elaboration of these problems, and implications for the 
conceptualization, are presented in the general discussion in Chapter 
7. 

The only other effect that was significant was a decision 
opportunity interaction with discrepancy (explaining about 1% of the 
variance); with the remaining F statistics each less than one. The 
best understanding of the data can be gained from a graphic 
representation (Figure 13). The overall pattern matches closely that 
which was hypothesized. The pattern for the (non-significant) 
interaction between involvement and discrepancy is reassuring. The 
unexpectedly significant interaction between decision and discrepancy 
once again suggests that making a decision makes subjects more critical 
of the product in their initial evaluation. (Hypothesis 6 proposed 
that a contrast effect would be evidenced, but in the same direction as 
the discrepancy.) This was true in the positive discrepancy conditions 
in which the unadulterated coffee was rated 5.62 by those subjects who 
had no decision opportunity but only 5.00 by those who decided on the 



286 



Combined 
Attitude 



7.0 p 



6.0 - 



5.0 



4.0 




3.0 



No decision 



— — • Low 

involvement 

High 

involvement 



Decision No decision 



Decision 



Positive 
discrepancy 



Negative 
discrepancy 



Figure 13 

Attitude Toward Coffee 

Mean Response by Experimental Condition 



287 

product. In the negative discrepancy conditions, the ratings among 
subjects not offered a decision, M = 4.55, and the ratings among those 
who did decide to try the coffee, M = 4.78, did not differ (a = .05, 
Tukey's HSD). 

Once again, there is mixed support for the overall 
conceptualization. While Hypothesis 3 does receive some statistical 
support (i.e., the non-significance of involvement or decision main 
effects), the resultant means indicate that there is a response to the 
manipulations. This suggests a confounding between attitude 
(Hypothesis 3) and satisfaction (Hypothesis 4) as currently measured in 
CS/D research. In light of this, it is not disquieting to find that 
involvement and decision did have some influence on the direction of 
effects. 

The six-item scale measuring attitude toward the coffeemaker also 
proved to be unidimensional . A principal components factor analysis 
revealed one factor, explaining 85% of the variance, which enabled use 
of a composite scale (Cronbach's a = .96, Table 47). As before, a 
lack of significant effects, as demonstrated by omnibus and effect F- 
statistics, was to be expected for the coffeemaker attitude measures. 
The results indicate this to be true, with subjects overall granting 
the coffeemakers a favorable, M = 5.30, rating (Table 48). 

Perception measures . One controversy in the CS/D literature has 
centered on the influence of perception on satisfaction. Because the 
dissertation distinguishes satisfaction from attitude, the effects of 
perception on each construct are of interest. 



288 



Table 47 
Reliability of Coffeemaker Attitude Scale 



a) Factor Analysis Results 3 

Factor 1 
.91 
.95 
.96 
.87 
.90 
.94 





Items 


1 


Good 


2 


Satisfactory 


3 


Favorable 


4 


Valuable 


5 


Like 


5 


Approve 



variance explained 85% 

b) Cronbach's Alpha Computation 13 

Pearson Correlation Coefficients: 

Item 3 ■ Item 4 Item 5 



.831 

.805 .726 

.890 .784 .842 





Item 1 


Item 


Item 2 


.848 




Item 3 


.831 


.932 


Item 4 


.708 


.784 


Item 5 


.841 


.791 


Item 6 


.796 


.909 



.96 



Note: a Principal component analysis. 

b Using formulation in Carmines and Zeller 1979. 



Table 48 
ANOVA Summary of Coffeemaker Attitude Ratings 



289 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



181 



8.57 
2.29 
3.49 
1 .22 
0.46 
1 .33 
1 .71 
0.27 
387.81 



Cell Means (variance) [n] 



0.57 
1 .07 
1 .63 
0.57 
0.22 
0.62 
0.80 
0.13 



Low Involvement, No Decision, Positive Discrepancy 5.35 (2.50) [25 

High Involvement, No Decision, Positive Discrepancy 5.25 (3.26) [25 

Low Involvement, No Decision, Negative Discrepancy 5.92 (0.70) [17 

High Involvement, No Decision, Negative Discrepancy 5.32 (1.28) [25 

Low Involvement, Decision, Positive Discrepancy 5.07 (2.71) [25 

High Involvement, Decision, Positive Discrepancy 5.19 (2.54) [24 

Low Involvement, Decision, Negative Discrepancy 5.28 (1.65) [25 

High Involvement, Decision, Negative Discrepancy 5.22 (2.03) [23 



Low Involvement 5.36 
High Involvement 5.24 



Main Effect Means 

No Decision 5.42 Positive Discrepancy 5.21 
Decision 5.19 Negative Discrepancy 5.40 



Note: Analysis of mean responses to combined seven-point semantic 
differential scales describing the respondent's feelings 
about the coffeemaker as bad/good, unsatisfactory/ 
satisfactory, unfavorable/favorable, worthless/valuable, 
dislike/like, disapprove/approve; scored one/seven 
respectively. Results are for a combined scale representing 
the average response, Cronbach's alpha = .97. Significance 
of results is indicated by: a = p< . 000 1 ; b = p< .01; c = 
p<.05; d = p< . 1 . 



290 
Hypothesis 7 proposed a direct relationship between perception of 
and attitude toward an object (i.e., a product such as the coffee). 
Such a simple relationship might be found in cases for which there is 
little other information on which to build, such as the trial of a new 
product. In the present study, subjects evaluated both a food product 
(coffee) and a mechanical product (cof feemaker) then rendered attitude 
judgments of each. 

For both types of product, correlations between perception 
measures and attitude measures were fairly strong (Table 49). While 
reassuring at first, there are some problems revealed in the table. 
The measures that showed no significant correlations related to the 
coffee's flavor strength and color. The other measures were phrased 
such that they could be interpreted as having evaluative dimensions 
analogous to attitude (although no true measures of evaluative 
dimension of attitude were collected). Indeed, regression analyses 
showed that two of the variables — poor/good quality, bad/good tasting — 
explained about 60% of the variance in attitude (no further 
interpretation of the regression analyses is warranted) . 

So although the hypothesis seems to be supported by the 
significant correlations, the findings can be questioned on the basis 
of confounding of perception and attitude. The criticism is blunted 
somewhat by the analysis of the coffeemaker data. Here, too, the 
correlations are significant; but the perception measures are less 
likely to be interpretable as direct measures of attitude. While some 
evaluative expression is present in the coffeemaker perception scales, 
the scales are more clearly attribute-specific. 



« a 



u-i m -u ^o ij ^o 



N m n N 



CI — I C"> CN 



— ( co <fl 

O £. £. 

a. a. 



03 


03 


a 


1) 


E 


e 


00 


en 


03 


03 



291 



c^ u"i r~- O 



TJ 


-a 


<U 


i> 


C 


c 






Uj 


UJ 


1) 


U 


-o 


■a 



i-l iJ 73 



• 




u O 01 


a 


l* -h "0 


^5 


CO • 3 


3 



1) \/ _J w 



.-i o < 



a u 

o < 



« 4) 41 

T3 m -u 4-i 

01 O -H IH 

4-i • a en 
u v o o 
o q. a. q. 
a BE 
V •• o o 
as o o u 



41 i-l >- <U 

.u > -* O -u 

u tg 1 h n 

•H -< 3 O (TJ 

pa c* c u h 



-* H 111 

a. aj a. 1- 



■-I 03 ■— J-l 
CO fc< CO < 



292 

In summary, better perception measures would have been desirable 
(e.g., those of Eastlack 1964, rather than Olson and Dover 1976 and 
Cohen and Goldberg 1970) and some measurement technology that would 
have avoided common method bias would have been preferred. 
Nevertheless, the hypothesis receives support, albeit from debatable 
evidence. 

Perception, itself, was expected to be affected by the extended 
conceptualization. That is, the early CS/D studies described in Table 
7 actually addressed performance perception. Not all, however, 
reflected the complete prepurchase, purchase, postpurchase sequence 
that is postulated to be necessary in assessing generalizable 
behavior. 

Hypothesis 6 stated that undertaking a decision process influences 
directly the perception of the outcome of the decision. So, in 
addition to the effect of the presence or absence of decision on 
satisfaction, there should be an effect on measures of discrepancy from 
expectations . 

The direction of this discrepancy was manipulated as part of the 
overall design, so there should have been an interaction between the 
effects of manipulated discrepancy and decision opportunity on 
perceived discrepancy. This problem of directionality can be 
alleviated by a number of different transformations; reported here are 
both raw and squared discrepancy means (Table 50). 

It was expected that the transformation would show a simple main 
effect of decision opportunity, so one-way Analyses of Variance 
(essentially, t-test-s) were performed and are reported in Table 50 for 



293 
both sets of dependent measures. The full ANOVA model was also applied 
to both sets of data when no significant simple effects were uncovered. 
As shown in the table, the only statistically significant influence on 
the dependent measures was the manipulated discrepancy. 

The simplest explanation for the failure to uncover a contrast 
effect is that it was not present. That is, either the discrepancies 
were perceived veridically or assimilation effects were present. This 
is consistent with reports of difficulty finding such an effect that 
date to the earliest days of CS/D research (e.g., Cardozo 1965, 
Olshavsky and Miller 1972). It is evident that quality differences 
were detected, so the assimilation was not as great as reported in some 
prior studies (e.g., Olson and Dover 1976, 1978). In summary, 
however, the analyses did not show that contrast did or could have 
occurred in this study. 

Satisfaction measures . If perception (and expectations) was 
manipulated in the desired direction, this should be reflected in the 
measures that are commonly used in CS/D studies to assess 
satisfaction. This is consistent with treatment under both the old and 
the extended conceptualizations. 

Hypothesis 2 states that commonly used measures of satisfaction 
respond to discrepancy between expectations and product performance 
even in the absence of a decision or adequate response involvement. 
One interpretation of this hypothesis is that only a main effect of 
discrepancy is predicted, with the positive discrepancy conditions 
netting a more positive satisfaction response. As depicted in Figure 
11b, however, an additional effect, an interaction between involvement 



294 



Table 50 
Significance Tests For Perceived Discrepancy Measures 

Squared Variables 

Measure e mean F, 194 ^7,137 mean F-j a 1 94 F 7j1 3 7 



1 . "Expectations too 

high/low" 4.1 0.02 2.26= 

2. "Coffee worse/better 

than expected" 4.6 2.65 1.45 



3. 


Mean of variables 
























1 and 2 


4.3 





68 


2 


00d 












4. 


Bitterness 


4.5 





11 





17 


1 .5 





62 


1 


39 


5. 


Flavor strength 


-.03 





91 





62 


4.0 





30 





74 


6. 


Quality 


5.2 





31 


1 


36 


1 .8 





64 


3 


72b 


7. 


Color 


3.2 


2 


69 


1 


27 


1 .3 





64 


4 


91« 


8. 


Taste 


5.8 


1 


81 





96 


1 .7 





61 


3 


00° 



Note: Non-response caused some minimal variation in actual degrees of 
freedom, but not enough to change the results. Statistical 
significance is indicated by a: p<.0001, b: p< . 01 , c: p<.05, d: 
p< .10. In each case, significance was attributable to a main 
effect of discrepancy condition, unencumbered by any 
interaction. 

e Discrepancy measures were described in more detail in the 

manipulation check discussion. Individual discrepancy scores 
were calculated for the coffee attributes as perception score 
minus expectation score. 



295 
and discrepancy, may obtain due to closer scrutiny of the product under 
high involvement conditions. The mean responses are drawn to 
illustrate a slight, albeit not statistically significant, such 
effect. 9 

Although allegedly measuring the same construct, the satisfaction 
scales are analyzed separately. This reflects the philosophical stand 
that the scales are being evaluated competitively (that is, as 
described in Chapter 3, no one scale is universally accepted among CS/D 
researchers). It is possible that there will be divergence in results 
among the scales. 10 

For this reason, the resultant means are displayed in Figure 14 
for each scale (in the interest of simplicity, the scales are referred 
to by labels consistent with Westbrook's 1983 usage: 14a Satisfaction; 
14b Odds; 14c D-T; 1 4d Faces and 14e Oliver). Although not perfect 
images of one another, the scales do exhibit a fairly high degree of 
similarity among themselves. The patterns also suggest some 
correspondence with that hypothesized in Figure 11b. 



9 It should be acknowledged at this point that possible weaknesses 
in the involvement and discrepancy manipulations render interpretation 
of results arguable. While an effect of discrepancy is expected, 
failure to find any effect of involvement could be dismissed as due to 
weakness in the manipulation. This problem has been discussed with the 
involvement manipulation check and further caveats are issued in the 
general discussion, Chapter 7. 

10 Obviously "consistency with hypotheses" is not an adequate 
criterion for selecting that satisfaction measure which best represents 
"truth". Control of measurement error and sensitivity are likely to be 
evidenced by F-tests for the effects within each measure. 



1 



296 



Satisfied 
7.0 r 



6.0 



5.0 




4.0 



3.0 I L 



Low involvement 
High involvement 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 

Figure 14 

Mean Responses to Standard Satisfaction Scales 

by Experimental Condition 

a) Satisfied Scale 



297 



Odds 
8.0 



7.0 



6.0 



5.0 



4.0 



3.0 




Low involvement 
High involvement 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 



Figure 14--continued 
b) Odds Scale 



298 



D-T 
7.0 r 



6.0 - 



5.0 



4.0 



3.0 




Low involvement 

■- High involvement 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 



Figure 14--cont inued 
c) Delighted-Terrible Scale 



299 



Faces 

7.0 r 



6.0 



5.0 



4.0 - 



Low involvement 

High involvement 



3.0 

No decision Decision No decision Decision 

Positive discrepancy Negative discrepancy 



Figure 14--continued 
d) Faces Scale 



n 



300 



Oliver 
7.0 r 



6.0 



5.0 



4.0 



3.0 



Low involvement 
High involvement 



No decision Decision No decision Decision 
Positive discrepancy Negative discrepancy 



Figure 14--continued 
e) Oliver's Likert-type Scale 



30" 
Analyses of variance indicate that significance of the results is 
less uniform, and therefore more ambiguous to interpret (Tables 51, 52, 
53, 54, and 55). Specifically, only the Odds measure demonstrated 
variability that was explained by the model at a conventional 
significance level (p < .10, Table 52). Total variance (sensitivity) 
for this measure is greater than that for the others due to the larger 
number of potential responses. Of the remaining scales, Faces (F 7]183 
= 1.59, p < .14, Table 54) and Oliver (F 7)186 = 1.52, p < .16, Table 
55) demonstrated effects that were better explained by the ANOVA model. 

Examining the results of analysis of each measure reveals 
interesting information. Using the Odds scale as the criterion, the 
expected discrepancy main effect was evidenced (accounting for 
approximately 3% of variance) and the interaction between discrepancy 
and involvement was in an appropriate pattern albeit not significant (p 
< .25). In addition to these effects was a significant (p < .10) 
response to the decision manipulation that indicated that the product 
was viewed more negatively in the conditions in which a decision was 
made (M = 5.45) than in those that allowed no decision (M = 6.15). 
Examination of the individually plotted means suggests that this was 
true for the positive discrepancy conditions but not the negative 
discrepancy conditions (using Tukey's HSD test, albeit in the face of a 
nonsignificant, p < .32, decision by discrepancy interaction). 
Overall, the data from the Odds scale seem to suggest that the 
manipulations did have their desired effects. Moreover, the Odds scale 
itself is more sensitive than predicted under Hypothesis 2. 



Table 51 
ANOVA Summary of Coffee Satisfaction Ratings 



302 



Source 



df 



Sums of Squares 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

BE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



1 . Low Involvement, 

2. High Involvement, 

3. Low Involvement, 

4. High Involvement, 

5. Low Involvement, 

6. High Involvement, 

7. Low Involvement, 

8. High Involvement, 



18.70 



0.74 



183 



0.05 
554.69 



0.88 
1 .03 
2.04 
0.86 
0.24 
1 .91 
0.20 
0.02 



Cell Means (variance) [n 



omega- 
squared 



No Decis 
No Decis 
No Decis 
No Decis 
Decision 
Decision 
Decision 
Decision 



on, 
on, 
on, 
on, 



Positive Discrepancy 5.50 (2.96) [24] 

Positive Discrepancy 5.76 (2.36) [25] 

Negative Discrepancy 5.50 (1.98) [22] 

Negative Discrepancy 5.00 (2.61) [24] 

Positive Discrepancy 5.16 (3.14) [24] 

Positive Discrepancy 5.13 (3.30) [23] 

Negative Discrepancy 5.35 (3.52) [26] 

Negative Discrepancy 4.68 (4.42) [22] 



Main Effect Means 



Low Involvement 5.37 
High Involvement 5.16 



No Decision 
Decision 



5.44 Positive Discrepancy 5.39 
5.09 Negative Discrepancy 5.14 



Note: Analysis of seven-point scale which asked, "How satisfied 

were you with the coffee?" Responses ranged from 1 = "Very 
Dissatisfied" to 7 = "Very Satisfied", including 4 = "Mixed" 
and = "Not Satisfied Nor Dissatisfied". Significance of 
results is indicated by: a = p<.0001; b = p<.01; c = p<.05; < 
= p<. 10. 



Table 52 
ANOVA Summary of Coffee Odds Scale 



303 



Source 



df 



Sums of Squares 



omega- 
sauared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



102.14 

57.00 

23.64 

1 .14 

7.95 

10.77 

0.83 

0.61 

1485.10 



Cell Means (variance) [n] 



85d 
22b 
,99d 

14 
01 
36 

10 



0.08 



.03 
.01 



Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 



Positive Discrepancy 6.56 (7.34) [25] 

Positive Discrepancy 7.20 (4.00) [25] 

Negative Discrepancy 5.43 (6.17) [23] 

Negative Discrepancy 5.36 (7.07) [25] 

Positive Discrepancy 5.48 (9.09) [25] 

Positive Discrepancy 6.08 (6.08) [24] 

Negative Discrepancy 5.38 (11.69)[26] 

Negative Discrepancy 4.83 (11. 70 ) [ 23 ] 



Main Effect Means 



Low Involvement 5.7 2 
High Involvement 5.89 



No Decision 
Decision 



6.15 
5.45 



Positive Discrepancy 6.33 
Negative Discrepancy 5.26 



Note : Analysis of eleven point scale rating the chances in ten the 
respondent "would choose to use the coffee again". Responses 
ranged from = "No Chance" to 10 = "Certain". See Appendix 
C for question wording and structure. Significance of 
results is indicated by: a = p< . 0001 ; b = p< . 01 ; c = p<.05; d 
= p< . 1 . 



Table 53 
ANOVA Summary of Coffee Delighted-Terrible Scale 



304 



Source 



df 



Sums of Squares 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



13.18 
4.96 
0.75 
0.02 
0.24 
1 .70 
0.39 
5.50 
334.64 



Cell Means (variance) [n 



1 .06 

2.79^ 

0.42 

0.01 

0. 14 

0.96 

0.22 

3.09^ 



omega- 
squared 



.01 



Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, No Decision, 

High Involvement, No Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 



Positive Discrepancy 5.32 (1.31) [25] 

Positive Discrepancy 5.24 (0.69) [25] 

Negative Discrepancy 4.78 (1.45) [23] 

Negative Discrepancy 5.00 (2.83) [25] 

Positive Discrepancy 4.88 (1.94) [25] 

Positive Discrepancy 5.29 (1.95) [24] 

Negative Discrepancy 5.15 (2.06) [26] 

Negative Discrepancy 4.52 (1.99) [23] 



Main Effect Means 



Low Involvement 5.04 
High Involvement 5.02 



No Decision 
Decision 



5.09 

4.97 



Positive Discrepancy 5.1. 
Negative Discrepancy 4.8. 



Note : Analysis of seven-point scale describing satisfaction with 
the coffee. Responses ranged from 1 = "Terrible" to 7 = 
"Delighted" and included 4 = "Mixed" and = "Neutral, Never 
Thought About It". See Appendix C for question and response 
wording. Significance of results is indicated by: a = 
p< . 0001 ; b = p< . 01 ; c = p<.05; d = p< . 1 . 



305 



Table 54 
ANOVA Summary of Coffee Faces Scale 



Source 



df 



Squares 


F 


omega- 
squared 


8.66 


1 .59 




5.61 


7.20b 


.03 


0.71 


0.91 




0.04 


0.05 




0.53 


0.68 




0.35 


0.46 




0.11 


0.15 




1 .38 


1 .77 





Model 

Discrepancy (DI) 
Decision (DE) 
Involvement (I) 
DE x DI 
I x DI 
I x DE 
I x DE x DI 
Residual 188 



146.30 



Cell Means (variance) [n_ 



1. Low Involvement, No Decision, Positive 

2. High Involvement, No Decision, Positive 

3. Low Involvement, No Decision, Negative 

4. High Involvement, No Decision, Negative 

5. Low Involvement, Decision, Positive 

6. High Involvement, Decision, Positive 

7. Low Involvement, Decision, Negative 

8. High Involvement, Decision, Negative 



Discrepancy 3.96 (0 

Discrepancy 3.80 (0 

Discrepancy 3.43 (0 

Discrepancy 3.44 (0 

Discrepancy 3.52 (0 

Discrepancy 3.79 (0 

Discrepancy 3.54 (0 

Discrepancy 3.30 (1 



87) [25' 

42) [25' 

62) [23" 

67) [25; 

76) [25; 

61) [24' 

82) [26' 

49) [23' 



Main Effect Means 



Low Involvement 3.62 
High Involvement 3.59 



No Decision 3.66 Positive Discrepancy 3.77 
Decision 3.54 Negative Discrepancy 3.43 



Note : Analysis of five point graphic scale rating "how you feel 
about the coffee". Responses were rescaled such that 1 = 
most negative face and 5 = most positive face. See Appendix 
C for wording and appearance of question. Significance of 
results is indicated by: a = p<.0001; b = p< . 01 ; c = p<.05; d 
= p<. 10. 



Table 55 
ANOVA Summary of Oliver's Six Item Scale 



305 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 
Decision (DE) 



Involvement 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



(I) 



186 



15.25 

1.52 
8.77 
0.25 
0.07 
3.75 
0.95 
0.01 
266.93 



Cell Means (variance) [n] 



1 .52 

1.06 

6.11 = 

0.18 

0.05 

2.61 

0.66 

0.00 



High Involvement, No Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 

Low Involvement, Decision, 

High Involvement, Decision, 



Main Effect Means 



Low Involvement 
High Involvement 



4.77 
4.84 



No Decision 
Decision 



4.59 
5.02 



03 



Low Involvement, No Decision, Positive Discrepancy 4.56 (0.72) [25] 
High Involvement, No Decision, Positive Discrepancy 4.76 (0.67) [25] 
Low Involvement, No Decision, Negative Discrepancy 4.69 (1.18) [23] 



Negative Discrepancy 4.35 (0.96) [24] 

Positive Discrepancy 4.87 (1.78) [25] 

Positive Discrepancy 5.38 (1.13) [24] 

Negative Discrepancy 4.95 (2.39) [25] 

Negative Discrepancy 4.87 (2.69) [23] 



Positive Discrepancy 4.89 
Negative Discrepancy 4.72 



Note : Analysis of mean responses to six seven-point Likert type 

items patterned after Oliver (1980): "I am satisfied about 
my choice of coffee; If I had it to do all over again, I 
would feel differently about choosing the coffee; My choice 
of coffee was a wise one; I feel bad about my decision 
regarding the coffee; I think that I did the right thing when 
I chose the coffee; I am not happy that I chose the coffee 
that I did. Mean responses are reported after reversing the 
First, Third, and Fifth scales so that 7 = Strongly Agree 
(with positive statements) and 1 = Strongly Disagree 
(Cronbach's alpha = .85). Significance of results is 
indicated by: a = p< . 0001 ; b = p < . 1 ; c = p<.05; d = p< . 1 . 



307 
The next most interesting pattern of responses were those to the 
D-T scale (Table 53). Since the diagrams in Figures 11 and 14 
effectively depict three-way breakdown of means, without a significant 
interaction hypothesized, it is useful to examine the one significant 
three-way interaction that did obtain (accounting for, however, only 
1% of variance). Pairwise comparisons (Tukey's HSD test) indicate that 
the means that were obtained for this measure, as shown in Figure 14c, 
are adequate representations of the pattern hypothesized and depicted 
in Figure 11b, with one additional complication. Apparently, the act 
of making a decision caused the divergence in effect of the involvement 
manipulation within the range of responses generated by the discrepancy 
manipulation. A significant main effect of discrepancy is also 
present indicating a more favorable evaluation of the product under 
positive discrepancy conditions, M = 5.18, than under negative 
discrepancy conditions, M = 4.88, omega-squared = .01. It is 
interesting to note that the obtained pattern is very much in accord 
with that predicted for the emotional measures of satisfaction and 
dissatisfaction (Figure 11a). Therefore, although Hypothesis 2 is not 
well supported by the DT results, Hypotheses 1a and 1b receive 
unexpected support. 

Of the three remaining scales, little needs to be said at this 
juncture. The analysis of variance in responses to the satisfaction 
scale revealed no significant effects (Table 51) and the main effects 
found to be significant for the Faces scale and Oliver's six-item scale 
(a = .85, Table 56) were consistent with the patterns already 
discussed. Mean responses to the Faces scale showed a more negative 



3( 

response to the negative discrepancy conditions, M = 3.43, than to the 
positive, M = 3.77 (accounting for approximately 3% of the variance in 
response to this 5-point scale). Oliver's scale seems to indicate that 
decision enhances satisfaction (M = 5.02) compared with no decision (M 
= 4.59) although this is questionable in light of no other effects 
(particularly, effects of discrepancy and its interactions). 

All in all, the patterns of results in Figure 14 display more 
sensitivity to the experimental manipulations than originally 
predicted. The first implication that can be drawn is that the 
manipulations, discrepancy and decision opportunity at least, were more 
effective than the check measures suggest. Secondly, some of the 
satisfaction measures may be more sensitive than may have been 
indicated by the global judgment offered earlier by the dissertation. 
This increases confidence regarding the manipulations and other 
analyses of the data. Further confidence is gained by examination of 
analogous scales for the coffeemaker product. In essence, only 
involvement (as the counterpoint to coffee involvement) was 
manipulated. Examination of the ANOVA tables for Satisfaction, Odds, 
D-T, and Faces scales evaluating the coffeemakers reveals no 
significant omnibus F-test (in fact, only one F of the four exceeded a 
value of one, Tables 57, 58, 59, 60). In light of strong expectations 
of no effect, these results are sufficient to judge the isolated 
statistically significant effects to be trivial (Tables 58, 60). It 
should be noted that satisfaction with the coffeemakers was generally 
good, but not outstanding (on the Satisfaction and D-T 7-point scales, 
means were 5.45 and 5.13, respectively; on the 11 -point Odds scale, 



H 



Table 56 
Reliability of Oliver's Six-item Scale 



a) Factor Analysis Results 3 

Itemsb Factor 1 

1 Satisfied with choice .74 

2 Feel differently again (-) .59 

3 Wise choice .77 

4 Feel bad (-) .77 

5 Did right thing .85 

6 Not happy (-) .83 



variance explained 58% 

b) Cronbach's Alpha Computation 

Pearson Correlation Coefficients: 

Item 3 Item 4 Item 5 



.435 

.674 .534 

.499 .742 .628 





Item 1 


Item 


Item 2 


.385 




Item 3 


.584 


.286 


Item 4 


.388 


.382 


Item 5 


.595 


.361 


Item 6 


.443 


.427 



Note : a Principal component analysis. 

b See Part VI of Exhibit 5 for full item statements. Items 

indicated by "(-)" were reversed for analysis. 
c Using formulation in Carmines and Zeller 1979. 



Table 57 
ANOVA Summary of Coffeemaker Satisfaction Ratings 



310 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



18.24 
1.52 
6.70 
6.48 
0.60 
0.03 
1 .30 
2.42 
586.35 



Cell Means (variance) [n] 



0.84 
0.49 
2.15 
2.08 
0.19 
0.01 
0.42 
0.77 



Low Involvement, No Decision, Positive Discrepancy 5.64 (3.16) [25. 

High Involvement, No Decision, Positive Discrepancy 5.36 (3.41) [25. 

Low Involvement, No Decision, Negative Discrepancy 6.17 (0.97) [23. 

High Involvement, No Decision, Negative Discrepancy 5.40 (3.58) [25. 

Low Involvement, Decision, Positive Discrepancy 5.44 (3.01) [25. 

High Involvement, Decision, Positive Discrepancy 5.04 (4.13) [24 

Low Involvement, Decision, Negative Discrepancy 5.31 (2.94) [26. 

High Involvement, Decision, Negative Discrepancy 5.30 (3.68) [23 

Main Effect Means 



Low Involvement 5.63 
High Involvement 5.28 



No Decision 5.63 Positive Discrepancy 5.37 
Decision 5.28 Negative Discrepancy 5.54 



Note : Analysis of seven-point scale which asked "How satisfied were 
you with the coffeemaker?" Responses ranged from 1 = "Very- 
Dissatisfied" to 7 = "Very Satisfied", including 4 = "Mixed" 
and = "Not Satisfied Nor Dissatisfied". Significance of 
results is indicated by: a = p< . 0001 ; b = p< . 01 ; c = p<.05; d 
= p< . 10. 



Table 58 
ANOVA Summary of Coffeemaker Odds Scale 



311 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 
Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



71 


44 





41 


18 


82 


27 


88 


13 


81 


2 


89 


5 


71 


2 


23 


1541 


84 



Cell Means (variance) [n. 



1 .24 

0.05 

2.29 

3.40d 

1.68 

0.35 

0.70 

0.27 



.01 



1. Low Involvement, No Decision, Positive Discrepancy 7.08 (5.66) [25] 

2. High Involvement, No Decision, Positive Discrepancy 6.44 (8.09) [25] 

3. Low Involvement, No Decision, Negative Discrepancy 6.91 (3.54) [22] 

4. High Involvement, No Decision, Negative Discrepancy 5.36 (9.66) [25] 

5. Low Involvement, Decision, Positive Discrepancy 5.80 (9.50) [25] 

6. High Involvement, Decision, Positive Discrepancy 5.42 (10.51 ) [ 24] 

7. Low Involvement, Decision, Negative Discrepancy 6.27 (10. 04) [ 26 ] 

8. High Involvement, Decision, Negative Discrepancy 5.83 (8.24) [23] 



Main Effect Means 



Low Involvement 6.51 
High Involvement 5.76 



No Decision 
Decision 



6.43 
5.84 



Positive Discrepancy 6.19 
Negative Discrepancy 6.08 



Note : Analysis of eleven point scale rating the chances in ten the 
respondent "would choose to use the coffeemaker again." 
Responses ranged from = "No Chance" to 10 = "Certain". See 
Appendix for question wording and structure. Significance of 
results is indicated by: a = p<.0001; b = p< .01; c = p<.05; d 
= p< . 1 . 



312 
mean response was 6.14 and on the 5-point Faces scale, mean response 
was 3.64). This is perfectly consistent with the discrepancy scale 
results reported earlier. 

Behavior measures . In addition to the need to debrief 
experimental participants, one purpose for the callback procedure was 
to collect self-reported behavioral data. This type of data collection 
has not been reported previously in the CS/D literature, but was 
thought to be a potentially rich source of information. Unfortunately, 
it proved to be more difficult to contact the subjects than had been 
anticipated. Of the 196 subjects who provided data used in analyzing 
the primary hypotheses, 83 were actually contacted one week after 
participating in the evaluation portion of the experiment. Since these 
contacts were approximately evenly distributed across conditions, the 
results are interpreted as though from a random sample, but with 
caution appropriate to the 42% response (Table 61). Hypotheses 4 and 
5 had been identified as secondary to the major contribution of the 
dissertation and the behavioral measures (self-reports) were collected, 
in part, to evaluate the feasibility of the methodology. 

Hypothesis 4 stated that satisfaction and dissatisfaction are each 
associated with a greater incidence of their respective theoretical 
outcomes — positive versus negative word-of-mouth, complimenting versus 
complaining--and that attitude is associated with intent to repurchase. 
Hypothesis 5 stated that the likelihood of behavior increases with 
strength of satisfaction or dissatisfaction emotion. Very few subjects 
reached criterion levels of satisfaction emotions (12) or 
dissatisfaction emotions (3), which rules out a test or discussion of 



Table 59 
ANOVA Summary of Coffeemaker Delighted-Terrible Scale 



313 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



187 



9.52 
1.58 
2.76 
1 .82 
1 .93 
0.27 
0.01 
1 .50 
329.02 



0.77 
0.90 



.57 
.04 
.10 
.15 



0.01 
0.85 



Cell Means (variance) [n_ 



Low Involvement, No Decision, Positive 

High Involvement, No Decision, Positive 

Low Involvement, No Decision, Negative 

High Involvement, No Decision, Negative 

Low Involvement, Decision, Positive 

High Involvement, Decision, Positive 

Low Involvement, Decision, Negative 

High Involvement, Decision, Negative 



Discrepancy 5.20 (1 .50) [25] 

Discrepancy 4.92 (1 .74) [25] 

Discrepancy 5.48 (0.53) [23] 

Discrepancy 5.40 (3.50) [25] 

Discrepancy 5.00 (1 .83) [25] 

Discrepancy 5.04 (1.78) [24] 

Discrepancy 5.23 (1.62) [26] 

Discrepancy 4.77 (1.42) [22] 



Low Involvement 5.22 
High Involvement 5.04 



Main Effect Means 

No Decision 5.24 Positive Discrepancy 5.04 
Decision 5.02 Negative Discrepancy 5.23 



Note : Analysis of seven-point scale describing satisfaction with 

the coffeemaker. Responses ranged from 1 = "Terrible" to 7 = 
"Delighted" and included 4 = "Mixed" and = "Neutral, Never 
Thought About It." See Appendix C for question and response 
wording. Significance of results is indicated by: a = 
p< .0001 ; b = p<.01; c = p<.05; d = p< . 1 . 



Table 60 
ANOVA Summary of Coffeemaker Faces Scale 



314 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



136 



5.38 
0.03 
0.28 
1 .31 
0.83 
0.17 
2.63 
0.18 
173.36 



Cell Means (variance) [n. 



0.82 

0.04 

0.30 

1 .41 

0.90 

0.19 

2.82«* 

0.20 



,01 



1. Low Involvement, No Decision, Positive Discrepancy 3.80 (1.00) [25] 

2. High Involvement, No Decision, Positive Discrepancy 3.40 (0.83) [25] 

3. Low Involvement, No Decision, Negative Discrepancy 3.96 (0.50) [23] 

4. High Involvement, No Decision, Negative Discrepancy 3.56 (1.09) [25] 

5. Low Involvement, Decision, Positive Discrepancy 3.56 (1.26) [25] 

6. High Involvement, Decision, Positive Discrepancy 3.75 (0.98) [24] 

7. Low Involvement, Decision, Negative Discrepancy 3.58 (0.89) [26] 

8. High Involvement, Decision, Negative Discrepancy 3.52 (0.86) [21] 

Main Effect Means 



Low Involvement 3.72 
High Involvement 3.56 



No Decision 3.67 Positive Discrepancy 3.63 
Decision 3.60 Negative Discrepancy 3.65 



Note: Analysis of five point graphic scale rating "how you feel 

about the coffeemaker". Responses were re-scaled such that 1 
= most negative face and 5 = most positive face. See 
Appendix C for wording and appearance of question. 
Significance of results is indicated by: a = p<.0001; b = 
p< . 01 ; c = p<.05; d = p< . 1 . 



Table 61 
Number of Subjects Contacted by Telephone 
at One Week Delay 



315 



Experimental Condition 



Number 
Number Not 
Contacted Contacted Total 



Low Involvement, No Decision, 
Positive Discrepancy 

High Involvement, No Decision, 
Positive Discrepancy 

Low Involvement, No Decision, 
Negative Discrepancy 

High Involvement, No Decision, 
Negative Discrepancy 

Low Involvement, Decision, 
Positive Discrepancy 

High Involvement, Decision, 
Positive Discrepancy 

Low Involvement, Decision, 
Negative Discrepancy 

High Involvement, Decision, 
Negative Discrepancy 



13 



11 



1 1 



11 



12 



83 



12 



14 

12 

18 

14 

14 

14 

15 
113 



25 



25 



23 



25 



25 



24 



26 



23 



316 
hypothesis 4 in its strictest interpretation (Table 66). Removing the 
restriction imposed by the criterion levels, and allowing satisfaction 
and dissatisfaction to vary along continua, collapses the essential 
points of hypotheses 4 and 5. Therefore, these are discussed together. 

The most straightforward assessment of the degree of association 
between variables is by correlation analysis. Measured levels of 
satisfaction, dissatisfaction and attitude toward the product were 
correlated (product-moment) with self-reported behavior or behavioral 
intention. 11 The results of these correlation analyses are reported 
here for the entire sample and for sub-samples blocked by discrepancy 
condition. The separate sample procedure allows satisfaction and 
dissatisfaction to follow different continua and generate--or respond 
to--separate psychological processes. 

Of greatest interest was the association between the emotional 
measures of satisfaction and dissatisfaction and their respective 
behavioral results. The satisfaction emotions shared no significant 
correlation with most of the behaviors for which effects were 
hypothesized — positive word of mouth, return of a complimenting 
postcard, consumption of the product, serving the product to others in 
the household or to guests — when examined either across the sample or 
just for those in the positive discrepancy conditions (Table 62 and 
Table 63, respectively). The only exception to this was a small 
positive relationship with a dummy-coded variable indicating whether or 
not positive word-of-mouth was reported (r=.22, p<.05). No such 



1 1 The selection of the gift product was a behavior observed by 
the experimenter. 



317 
relationship was found with the measure of the extent of word-of-mouth 
(i.e., the number of people spoken to positively). The more 
conventional measures of satisfaction fared a little better in the full 
sample, with significant correlations with the incidence and frequency 
of positive or negative word-of-mouth. These relationships appear to 
have been much stronger for those subjects in the negative discrepancy 
conditions (Table 64) than in the positive discrepancy conditions 
(Table 63), although the interpretations must be tempered in light of 
the low recontact rates. 

Similar discouraging results were encountered for the emotional 
measures of dissatisfaction and the reports of negative behaviors 
(Table 62). The dissatisfaction measure exhibited a slight 
relationship with the return of a postcard with complaints (r=.12, 
p< .10) as well as inverse relationships with the return of a 
complimenting card (r=-.13, p< . 10) and serving the coffee to guests 
(r=-.29, p<.05). Contrary to theory, however, was a direct 
relationship with engaging in positive word-of-mouth (r=.21, p< . 10) . 
Again, the results were mirrored in the within discrepancy analyses, 
with the positive discrepancy conditions apparently accounting for the 
correlation between the positive word-of-mouth and dissatisfaction 
emotion measures (Tables 63 and 64). 12 



12 Similar correlation analyses were conducted within other 
subsets of the data. Dividing the data set at either the midpoint of 
each satisfaction scale or at the median response to each scale did not 
affect markedly the interpretability of the significant correlations. 
Cross-tabulations and/or tests of differences between means were 
similarly unenlightening. Consequently, none of these analyses are 
reported nor discussed here. 



•1-1 . , 



318 



.* 6 



co co co 
CN i-l Ov 
Xi CX5 r-» 



en i—i 



£ 5 



5 * 






« CD 



I — H 



00 r>» i-i —i 



to 


— i 


IB 




o 


V 


-c 


o 


1 — 1 


01 


o 


CO 


-ui 




CJ 


Vj 


V 


CO 





a. 





r4 JZ 

CU J-J 

> -H 

■h 3 



u cu en >-i 01 co 

»< £H O 9 E h co 

CO 4J 0. CL. (J E OJ-O-U 

o a o •— • •" ox:— i en 



OO C t) ^ u 3 O 

"O I 01 4) J-i O -u 

hi -o ti j! c-^-as 

00 - D O i- en ai cu 

co o.a a 3 o > c > 



s a, o o 



o o c -a -h 

-a i cu cu jj 

01 T3 1- J"! C CO 

SO h V O h 01 

a) O -Q Q. 3 0) 



o t-> -a -h 

O -u en 3 >■< 

_, II u 341 

"O aj co .O o> 

>> 0) > 

U -i-l -O SO u u 

j-i i-i cu c oi o< 

U U -H fl CO 

-o a )- -i-i 

— < 0) 01 3 CO i—i 

3 > — i "O C — i 



319 



CO < _ | 



um > 



c J - 



5* 



« « 



o ^ 2 



CJ "*■ " 

w 4-1 






™ «W 



0) 


























3 


J3 


u 










CO 


-Q 






CO 


co co co 


4-1 


O 


lf"l 










in 


<r 






LH 


o cn n 




u-i 


m 










vO 


un 






nO 


sO 00 p- 


4-1 


























4-1 














i 












< 


























CO IM 

4) 4) 




"0 




















•O JS J3 


(0 -4 




00 




















r~- \0 on 




CN 




















i— 1 CN CN 


CJ i-l 


























CO O 


























c 

CO 


_□ 


"0 










CO 


J2 






CO 


co co co 


■H 0) 


n£> 


ON 










O 


LO 






CN 


OO O — i 


4J CJ 


<r 


CN 










vD 


<r 






vO 


^ rv vO 


U CO 


























eg b, 














1 


1 










en 


J3 


u 










CO 


JC 






CO 


CO co CO 


•H H 


co 


00 










nO 


<r 






ON 


ON O c-i 


4-1 1 


•* 


c-i 










u-i 


<r 






u-i 


-3* I""- nO 


CO Q 


























CO 














1 


i 










co 


CO 


j3 






-o 




CO 


CJ 






CO 


co co co 


"3 


nO 








m 




U3 


00 






ON 


ON r. cn 


"O 


LA 


<f 






CI 




u-| 


cn 






UN 


<• r-~ r~ 


o 














1 


1 










-a 


























oi 


_2 


CJ 










A 


-o 




J 


J3 


co co co 


UH 


m 


00 










ON 


oo 




ON 


CN 


nO nO On 


00 


u-l 


CI 










<r 


CN 




CN 


CO 


,<r \o un 


■-H 


























4-1 














i 


1 




1 






CO 


























w 


























c 


























c 


























•1- 


























JJ 


























u 


























<0 


T3 


CJ 






^3 














14-1 


CN 










<r 














CO 


c-> 


CN 








c-> 














•,M 


























u 












1 














cO 


























.. tn 


























co co 


























C "H 


























O Q 




















































4-1 


























c 


























E O 


























Ed —i 


























u 


o 


u 








■o 




-a 






■o 


u -a 


CO 


m 














r-s 






ON 


O On 


LH 


en 


m 








ri 




CN 






CN 


CN —l 


CO 






































I 














4J 


























CO 


























CO 




























CN 


CN 




ON 


CN 


CN 


CN 


CN 




CN 


<J CN 


m on on 


c 


<T 


01 

> 


4-1 
1"( 


O 


CI 


c-> 


<r 


0) 

> 


—i 


CT> 


01 
01 CO 
CO co 


ON ON ON 




<u 


■w 


3 








01 


•^ 


3 




co JS 


-a 




p» 


4-1 




4-1 






> 


4-1 




4-1 


-C O 


c 




■i-4 


■rJ 


T3 


c 








CO 


■a 


c 


a u 


CO 




4-1 


01 CO 


1m 


01 


CO 




-j 


oi oo 


u 


CU 


U 3 


u 01 




■H £ 


— i o 


CO 




l-l 


CO 


CO £ 


— I 01 


CO 


E 


3 a. Xi oo 




CO 4-J 


o. o. 


O 


g 


01 -o 


4J 


4-1 


o- c 


CJ 


S 


a 


4>N C 




3 


o •— - 


4-1 


o 


SI i-l 


CO 


4) 3 


o ^ 


-J 





o 


4-1 "O -H 




Cu O 


01 


CO 


o 


4J O 


0) 


Z 


oi 


CO 


O 


4-1 


CO 3 >N 






Sm o 







O JS 


3 


X 


Cm 







4-1 


CU 4-1 3 V 




C i 


H 


04 


0) 


01 


CJ 


C I 


i- 


Cm 


01 


T3 


4J IB J (1 


CO 


■_ 1 C4-I 


UM 




> 


O CD 




1-1 U-I 


U-4 




> 


>N CU 


> 


u 


O 


c 


-a 


■H 


4-1 3 





O 


C 


-o 


■-^ 


U 'H 


-O 00 1m 1m 





"O 1 


01 


0) 


4-1 


O 


4-1 


-a i 


01 


1) 


4-1 


4-1 J- 


01 C 01 01 




0) "O 


U J* 


c 


1-1 


T3 =C 




oi -o 


u J* 


C 


CO 


4-1 


4J -l 13 CO 


> 


OC 1- 


01 o 


Ij 


CO 


01 


01 


oo u 


0) 


lM 


SO 


■o 


O U -H 


CO 


co 


jd a. 


a 


o 


> C 


> 


<o o 


x> a. 


3 


01 


-M 01 


« 3 H H 


x 


00 3 


B co 


4-) 


Oh 


>-l -H 


1-1 


00 3 


e co 


4-1 


z 


3 > 


MM -O C l-H 


01 


c 


3 


0) 




0) 


01 


c 


3 


0) 




O CO 


0) O -H 


ca 


w 


2 


o£ 




CO 


CO 


w 


Z 


cc 




3 33 


CO cj 3 



CJ O > 
3 • CO 
■O V -3 



ti u 



05 < _ 1 



to « 



° 4J 



S 3 <" 



z a ° 



h x n n ri 



320 



S u > 
o < £ 



— i O T3 "0 



5 C 



* u-, 



O 4J 
■o O -i 
41 O Kl 



41 05 

en co 

« .c -a 



■h -a c 

4> 05 l~ 4> w 

— I O cfl § I- 

a a. o B 4) 



(0O^5CU3O>C> 

003 E c/5 4-> a. t-i-H u 



co x: — ' oj <fl 



s a- o o 



O O C T3 -I 

"O I 4) 0) *J 

» 13 h J! C fl 

5t I. -J O U JC 

co XI Q. 3 4) 

00 3 E W 4-1 Z 






4-1 


05 


3 


>, 




-J 




4) 


4-1 


3 


41 




-o 


4-1 


05 


JO 


41 


:* 


11 








> 


h 




T) 


SO 


u 


U 


4-1 


U 


41 


c 


4) 


4) 




4-1 




■-J 


■o 


w 


■a 




CJ 


U 


•H 




t— * 


41 


41 


3 


05 


fi 


3 


> 




-3 


C 







CO 


41 









3 


— 


W5 




u 


LS 



•W T3 4) 

4-i 4i a 

eg 4J -H 



321 
It is interesting to note that the satisfaction measures adapted 
from prior research did, overall, correlate more highly with the target 
behaviors. However, the measure used for attitude toward the coffee 
also varied with the behaviors. This indicates that — at least at the 
time of measurement — the constructs are highly confounded, using such 
similar assessment instruments. 

Further evidence of the time-sensitive nature of the processes can 
be seen in the relative magnitudes of the correlations. The strongest 
correlations were among the satisfaction and attitude measures with the 
behavioral intent measures that were collected at the same time, as 
well as the same affective or cognitive measures correlated with gift 
selection behavior. Other correlations, collected after a delay and 
possible reconsideration of the product, were less strong. Also, 
consistent with the Theory of Reasoned Action, the behavioral intention 
measures were generally more highly related to attitude — and the 
satisfaction measures — than were the reports of actual behaviors. 

In summary, the conclusions that can be drawn from the behavioral 
assessment procedures are tentative at best. The rate of re-contact 
was too low to inspire confidence and netted very small sample sizes in 
some blocked analyses. Further, those analyses that depended on the 
occurrence of relatively extreme behaviors (e.g., return of a self- 
addressed, stamped postcard) would require much larger overall samples, 
as these behaviors proved to be rare events. While hypotheses 
regarding such events are testable comparing against appropriate 
statistical distributions, the results of the present study do not 
display effects that warrant such treatment. 



322 

The logic behind the hypotheses and analyses of behavioral 
results, however, remains appealing. The CS/D literature is rooted in 
an interest in postpurchase behaviors. As discussed previously, one 
difficulty in finding behavioral effects in CS/D research arises from 
the interference of constraints on behavior. The preceding analysis 
attempted to control these extraneous factors at a constant level. 
Another approach would be to evaluate a construct, such as motivation 
to behave, which acts earlier in the process. 

Hypothesis 8 makes explicit the "conservation of motivation" 
concept elaborated earlier. It states that strength of emotional 
response varies with involvement. Again, correlation analysis is 
appropriate to the question. 

Since response involvement was both manipulated and measured, it 
was possible to analyze the effect on emotions (particularly those 
hypothesized to be characteristic of satisfaction feelings) of assigned 
condition and perceived involvement. The former was evaluated by means 
tests and the latter by correlation. 

There were no significant differences in mean emotion strength 
(individual emotions) detectable by either t-test or ANOVA (details of 
which are not reported here. There were, however, some modest 
correlations between measures of involvement and emotions for the 
overall sample as well as subsets separated by discrepancy condition 
(Table 65). None of the correlations was very large, however, 
suggesting very limited support of the hypothesis. 

As noted in the manipulation check discussion, the response 
involvement manipulation was the most equivocal and the measures of 



323 



Table 65 
Correlations Between Involvement and Emotion Measures 



a) Full data set (n=188) 

Involvement e 
Emotions f Interesting Important Difficult Involving Evaluation 



Satisfaction 
Dissatisfaction 


-.12 d 


Interest 


.15 c 


Joy 




Surprise 
Anger 


-.18 c 


Disgust 




Distress 


-.17 c 


Contempt 


-.14 c 


Fear 




Shame 
Guilt 


-.19 c 
-.14 d 



25 c 



16 c .20 b 



.24 b -.13 d .19 c .23 b 
.28 a -12 d -16 c 



.13 d 

.14 d .16 c 

13 d 



b) Data segregated by discrepancy conditions 



Interesting Important Difficult Involving Evaluation 

18 d .21 c .30 b -32 b 

.22 c 
20 d .19 d .31 b -23 c .21 b -26 c 

20 d .25 c .30 b -21 c 

.26 c 

.17 d 

.27 b 
.23 c .20 c -21 c 



Satisfaction 




Dissatis- 
faction 


-.21 c 


Interest 




Joy 




Surprise 
Anger 


-.27 b 


Disgust 
Distress 


-.17 d 


Contempt 


-.24 c 


Fear 
Shame 


-.26 b 


Guilt 





.24< 



19 



Note to Table 65: : Reported are Pearson correlations; a: p<.0001, b 

p< .01 , c: p<.05, d: p< . 1 . 
e Involvement measures as defined previously. 
f Emotion measures as defined previously. 
g Reported are: first positive n = (99) then negative (n = 90) 

discrepancy within each pair. 



324 



52 5 



involvement were thought to be unreliable. This may reflect a floor 
effect in which the chosen situation was unable to adequately distract 
a subject from an inherently uninvolving product. Again, at best only 
tenuous support can be derived for the hypothesis. 

Measures based on the Differential Emotions Scale . The most 
direct assessment of the influence of the constructs added by the 
conceptualization on satisfaction, as defined, is by analysis of the 
DES patterns of emotions. Prediction of effects measurable by this 
technique was the focus of the hypotheses central to the contribution 
of the extended conceptualization. 

Hypotheses 1a and 1b stated that, in the presence of all enabling 
conditions, a positive discrepancy from expectations to performance 
perception results in a pattern of emotions that constitutes 
satisfaction and a negative discrepancy results in a pattern of 
emotions that constitutes dissatisfaction. The pilot study reported in 
Chapter 5 concluded that appropriate definitions of these patterns of 
emotions required average ratings of 3 on the Differential Emotions 
Scale items which described one's feelings of interest, joy, and 
surprise (satisfaction) or anger, disgust and surprise 
(dissatisfaction) . 

This measurement device was quite disappointing for the present 
study. An examination of responses by condition experienced reveals no 
mean equal to nor exceeding 3.0 (Tables 67 and 68). An inspection of 
the data at the level of individual subject response shows that few 
subjects ever reached the criterion level of satisfaction or 
dissatisfaction (Table 66). 



1 



326 



Table 66 
Number of Subjects Reaching Criterion Emotion Levels 



Experimental Treatments 



Number (%) At or Above Criterion 
Satisfaction Dissatisfaction 



Low Involvement, No Decision, 

Positive Discrepancy 
High Involvement, No Decision, 

Positive Discrepancy- 
Low Involvement, No Decision, 

Negative Discrepancy 
High Involvement, No Decision, 

Negative Discrepancy 
Low Involvement, Decision, 

Positive Discrepancy 
High Involvement, Decision, 

Positive Discrepancy 
Low Involvement, Decision, 

Negative Discrepancy 
High Involvement, Decision, 

Negative Discrepancy 



3 (12) 


(0) 


1 (4) 


(0) 


(0) 


(0) 


2 (8) 


1 (4) 


2 (8) 


1 (4) 


1 (4) 


(0) 


1 (4) 


(0) 


2 (9) 


1 (4) 



Table 67 
ANOVA Summary of Satisfaction Emotions 



327 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



7 
1 
1 
1 
1 
1 
1 
1 
187 



4.70 
0.18 
3.13 
0.37 
0.00 
0.15 
0.29 
0.58 
94.66 



1.33 

0.35 

6.18= 

0.74 

0.01 

0.30 

0.57 

1 .14 



Cell Means (variance) [n] 



.03 



Low Involvement, No Decision, Positive Discrepancy 1.88 (0.71) [25] 

High Involvement, No Decision, Positive Discrepancy 1.70 (0.52) [25] 

Low Involvement, No Decision, Negative Discrepancy 1.64 (0.35) [22] 

High Involvement, No Decision, Negative Discrepancy 1.80 (0.39) [25] 

Low Involvement, Decision, Positive Discrepancy 2.09 (0.71) [25] 

High Involvement, Decision, Positive Discrepancy 1.98 (0.45) [24] 

Low Involvement, Decision, Negative Discrepancy 2.09 (0.44) [26] 

High Involvement, Decision, Negative Discrepancy 1.87 (0.45) [23] 



Low Involvement 1.93 
High Involvement 1.83 



Main Effect Means 

No Decision 1.76 Positive Discrepancy 1.91 
Decision 2.01 Negative Discrepancy 1.86 



Note : Analysis of mean responses to nine Differential Emotions 

Scale items which were defined as comprising the satisfaction 
pattern of emotions: attentive, concentrating, alert, 
delighted, happy, joyful, surprised, amazed, astonished; 
Cronbach's alpha = .85. Significance of results is indicated 
by: a = p < . 6 1 ; b = p< . 1 ; c = p<.05; d = p< . 1 . 



Table 68 
ANOVA Summary of Dissatisfaction Emotions 



328 



Source 



df 



Sums of Squares 



Model 

Discrepancy (DI) 
Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



1 . Low Involvement, 

2. High Involvement, 

3. Low Involvement, 

4. High Involvement, 

5. Low Involvement, 

6. High Involvement, 

7. Low Involvement, 

8. High Involvement, 



1 .88 
0.01 
1 .01 
0.21 
0.09 
0.23 
0.11 
0.20 
38.16 



Cell Means (variance) [n] 



No Decision, 
No Decision, 
No Decision, 
No Decision, 
Decision, 
Decision, 
Decision, 
Decision, 



Positive 
Positive 
Negative 
Negative 
Positive 
Positive 
Negative 
Negative 



Discrepancy 
Discrepancy 
Discrepancy 
Discrepancy 
Discrepancy 
Discrepancy 
Discrepancy 
Discrepancy 



F omega- 
squared 



1 .32 

0.03 
4.98= 
1 .02 
0.44 
1 .12 
0.53 
1 .01 



,02 



1 .28 


(0.20) 


[25] 


1 .13 


(0.04) 


[25J 


1 .18 


(0.06) 


[23J 


1 .29 


(0.28) 


[25] 


1 .45 


(0.36) 


[25] 


1 .33 


(0.24) 


[24] 


1.39 


(0.21) 


[26] 


1.28 


(0.22) 


[23] 



Low Involvement 1.33 
High Involvement 1.26 



Main Effect Means 

No Decision 1.22 Positive Discrepancy 1.30 
Decision 1.37 Negative Discrepancy 1.29 



Note : Analysis of mean responses to nine Differential Emotion Scale 
items which were defined as comprising the dissatisfaction 
pattern of emotions: enraged, angry, mad, feeling of 
distaste, feeling of revulsion, disgusted, surprised, amazed, 
astonished; Cronbach's alpha = .80. Significance of results 
is indicated by: a = p< . 0001 ; b = p< .01 ; c = p<.05; d = 
p<. 10. 



329 
Analyses of variance were conducted on each composite scale. 13 
Neither scale showed the expected interaction among the three 
manipulated factors. The only significant effect on either measure was 
a main effect of the decision manipulation (Tables 67, 68). In each 
case, making a decision appears to have enhanced emotional response. 
For the satisfaction measure, mean responses were 1.76 for the no 
decision conditions versus 2.01 for the decision conditions (accounting 
for about 3% of the variance). Mean responses to the dissatisfaction 
measure were 1.22 in the no decision conditions and 1.37 in the 
decision conditions (accounting for about 2% of the variance). In 
light of the non-significant omnibus F-tests and the relatively slight 
explanatory power, even these results must be viewed with caution. 

One diagnostic strategy of interest is the analysis of each 
component emotion. Each emotion was measured using three items from 
the DES (see Table 69 for calculation of Cronbach alphas) and could 
also be expected to show response to the manipulations. Of the five 
emotions, none exhibited a significant omnibus F indicating 
statistically significant variation, although some effects — interpreted 
as a priori — appeared significant for each emotion except interest 
(Table 70). 

The effects witnessed on the satisfaction and dissatisfaction 
scales are reflections of responses to the joy and surprise scales 
(satisfaction) and the anger and surprise scales (dissatisfaction), all 



13 As discussed in Chapter 5, the component emotions are not 
expected to correlate directly so an internal reliability coefficient- 
such as Cronbach' s alpha--is less meaningful for these data. 
Nevertheless, values of .85 and .90 were calculated for satisfaction 
and dissatisfaction, respectively, and are reported. 



330 



Table 69 
Reliability of Emotion Measures 



a) Factor Analysis Results 3 

Basic Measurement Factor Factor Factor 

Item Item 1 2 3 

Interest Attentive (E1) .81 

Concentrating (E2) .72 
Alert (E3) " .74 

Joy Delighted (E4) .77 

Happy (E5) .77 

Joyful (E6) .63 

Surprise Surprised (E7) .61 

Amazed (E8) .67 

Astonished (E9) .74 

Disgust Feeling of Distaste (E10) .84 

Disgusted (E1 1 ) .77 

Feeling of Revulsion (E12) .70 

Anger Enraged (E13) -80 

Angry (E14) .28 

Mad (E15) .56 



variance explained 24.6% 19.5% 1 3 . 8%/ 5 7 . . 



331 



Table 69 — continued 

b) Cronbach's Alpha Computations For Each Emotion 13 
Pearson Correlation Coefficients 

attentive concentrating 
concentrating .54 a = .80 (Interest) 



a = .80 (Joy) 



alert 


.60 


.56 




delighted 


happy 


happy- 


.64 




joyful 


.56 


.51 




surprised 


amazed 


amazed 


.53 




astonished 


.61 


.71 




F. distaste 


disgusted 


disgusted 


.55 




F. revulsion 


.27 


.17 




enraged 


angry 


angry 


.26 




mad 


.30 


.11 



a = .83 (Surprise) 



a = .60 (Disgust) 



a = .46 (Anger) 



1 



332 
Table 69 — continued 

c) Cronbach's Alpha Computations for Satisfaction/Dissatisfaction 
Patterns 

Pearson Correlation Coefficients — Satisfaction 
E1 E2 E3 E4 E5 E6 E7 E8 



E2 


.54 














E3 


.60 


.56 












E4 


.47 


.42 


.51 










E5 


.49 


.50 


.46 


.64 








E6 


.35 


.32 


.33 


.56 


.51 






E7 


.25 


.44 


.41 


.30 


.33 


.24 




E8 


.03 


.28 


.30 


.29 


.21 


.30 


.53 


E9 


.04 


.26 


.30 


.26 


.28 


.33 


.61 



71 



Pearson Correlation Coefficients — Dissatisfaction 
E7 E8 E9 E10 E1 1 E12 E13 E14 



E8 


.53 














E9 


.61 


.71 












E10 


.47 


.39 


.59 










E11 


.22 


.43 


.46 


.55 








E12 


.36 


.35 


.46 


.27 


.17 






E13 


.36 


.41 


.48 


.16 


.16 


.60 




E14 


.18 


.27 


.20 


.16 


.21 


.18 


.26 


E15 


.05 


.09 


.22 


. 11 


. 1 1 


.11 


.30 



1 1 



Note: a Principal component analysis using varimax rotation. 
b Using formulation in Carmines and Zeller 1979. 



Table 70 
ANOVA Summary of Interest 



333 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



187 



4.76 
0.00 
1 .65 
0.00 

0.04 
0.27 
1 .26 
1 .63 
194.56 



0.65 
0.00 
1 .59 
0.00 
0.04 
0.26 
1 .21 
1 .57 



Cell Means (variance) [n. 



Low Involvement, No Decision, Positive Discrepancy 2.20 (1.51) [25] 

High Involvement, No Decision, Positive Discrepancy 2.10 (1.17) [25] 

Low Involvement, No Decision, Negative Discrepancy 1.92 (0.69) [22] 

High Involvement, No Decision, Negative Discrepancy 2.33 (1.02) [25] 

Low Involvement, Decision, Positive Discrepancy 2.33 (0.87) [25] 

High Involvement, Decision, Positive Discrepancy 2.28 (0.94) [24] 

Low Involvement, Decision, Negative Discrepancy 2.47 (0.91) [26] 

High Involvement, Decision, Negative Discrepancy 2.20 (1.18) [23] 

Main Effect Means 



Low Involvement 2.24 
High Involvement 2.23 



No Decision 2.14 Positive Discrepancy 2.23 
Decision 2.33 Negative Discrepancy 2.24 



Note : Analysis of mean responses to three Differential Emotions 
Scale items which were defined as comprising the interest 
emotion: attentive, concentrating, alert; Cronbach's alpha = 
.80. Significance of results is indicated by: a = p<.0001; b 
= p< . 01 ; c = p<.05; d = p< . 1 . 



Table 71 
ANOVA Summary of Joy 



334 



Source 



df 



Sums of Squares 



F omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 
Low Involvement, 
High Involvement, 



7 
1 
1 
1 
1 
1 
1 
1 
187 



38 
69 
15 
25 



,63 
,31 
,03b 

.71 



.03 



0.01 
0.00 
0.01 
0.17 
136.97 



0.01 
0.00 
0.02 
0.23 



Cell Means (variance) [n] 

No Decision, Positive Discrepancy 1.91 (0.88) [25. 

No Decision, Positive Discrepancy 1.68 (0.74) [25 

No Decision, Negative Discrepancy 1.66 (0.63) [22 

No Decision, Negative Discrepancy 1.53 (0.53) [25. 

Decision, Positive Discrepancy 2.15 (1.23) [25 

Decision, Positive Discrepancy 2.07 (0.73) [24 

Decision, Negative Discrepancy 2.04 (0.42) [26 

Decision, Negative Discrepancy 1.83 (0.71) [23. 

Main Effect Means 



Low Involvement 1.95 
High Involvement 1.77 



No Decision 
Decision 



1 .69 
2.02 



Positive Discrepancy 1 
Negative Discrepancy 1 



.95 

,77 



Note: Analysis of mean responses to three Differential Emotions 

Scale items which were defined as comprising the joy emotion: 
delighted, happy, joyful; Cronbach's alpha = .80. 
Significance of results is indicated by: a = p<.0001; b = 
p< . 01 ; c = p<.05; 
d = p<. 10. 



Table 72 
ANOVA Summary of Surprise 



335 



Source 



df 



Sums of Squares 



F omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



4.82 
0.00 
3.07 



187 






49 





01 





50 





36 





35 


17 


89 



1 .09 

0.00 

4.87c 

0.78 

0.01 

0.79 

0.59 

0.55 



,02 



1. Low Involvement, 

2. High Involvement, 

3. Low Involvement, 

4. High Involvement, 

5. Low Involvement, 

6. High Involvement, 

7. Low Involvement, 

8. High Involvement, 



Cell Means (variance) [n] 

No Decision, Positive Discrepancy 1.52 (0.56) [25] 

No Decision, Positive Discrepancy 1.32 (0.30) [25] 

No Decision, Negative Discrepancy 1.35 (0.28) [22] 

No Decision, Negative Discrepancy 1.52 (0.58) [25] 

Decision, Positive Discrepancy 1.79 (1.21) [25] 

Decision, Positive Discrepancy 1.58 (0.63) [24] 

Decision, Negative Discrepancy 1.76 (0.66) [26] 

Decision, Negative Discrepancy 1.59 (0.79) [23] 

Main Effect Means 



Low Involvement 1.61 
High Involvement 1.50 



No Decision 1.43 Positive Discrepancy 1.55 
Decision 1.68 Negative Discrepancy 1.56 



Note : Analysis of mean responses to three Differential Emotions 
Scale items which were defined as comprising the surprise 
emotion: surprised, amazed, astonished; Cronbach's alpha = 
.83. Significance of results is indicated by: a = p< . 0001 ; 
= p< .01 ; c = p<.05; d = p<. 10. 



Table 73 
ANOVA Summary of Anger 



336 



Source 



df 



Sums of Squares 



omega- 
squared 



Model 

Discrepancy (DI) 

Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



1 . Low Involvement, 

2. High Involvement, 

3. Low Involvement, 

4. High Involvement, 

5. Low Involvement, 

6. High Involvement, 

7. Low Involvement, 

8. High Involvement, 



2.01 
0.07 
0.92 
0.43 
0.14 
0.05 
0.13 
0.23 
33.50 



Cell Means (variance) [n] 



No Decision, 
No Decision, 
No Decision, 
No Decision, 
Decision, 
Decision, 
Decision, 
Decision, 



Positive 
Positive 
Negative 
Negative 
Positive 
Positive 
Negative 
Negative 



Disc 
Disc 
Disc 
Disc 
Disc 
Disc 
Disc 
Disc 



repancy 
repancy 
repancy 
repancy 
repancy 
repancy 
repancy 
repancy 



Main Effect Means 



1.61 

0.40 

5.15= 

2.44 

0.81 

0.28 

0.72 

1.31 



.09 
.01 
.07 
.07 
.27 
.22 
.28 
.03 



(0.11) 
(0.00) 
(0.06) 
(0.06) 
(0.3C) 
(0.45) 
(0.41) 
(0.02) 



Low Involvement 1.18 
High Involvement 1.08 



No Decision 
Decision 



1 .06 
1.20 



02 



[25] 
[25] 
[23] 
[25] 
[25] 
[24] 
[26] 
T23] 



Positive Discrepancy 1.15 
Negative Discrepancy 1.12 



Note : Analysis of mean responses to three Differential Emotions 
Scale items which were defined as comprising the anger 
emotion: enraged, angry, mad; Cronbach's alpha = .46. 
Significance of results is indicated by: a = p< . 000 1 ; b = 
p< . 01 ; c = p<.05; d = p< . 1 . 



Table 74 
ANOVA Summary of Disgust 



337 



Source 



Model 

Discrepancy (DI) 
Decision (DE) 

Involvement (I) 

DE x DI 

I x DI 

I x DE 

I x DE x DI 

Residual 



df 



Sums of Squares 



7 
1 
1 
1 
1 
1 
1 
1 
187 



1 . 19 
0.00 
0.07 
0.00 
0.21 
0.82 
0.00 
0.06 
41 .38 



Cell Means (variance) [n] 



omega- 
squared 



0.77 
0.02 
0.31 
0.01 

0.96 
3.71^ 
0.00 
0.27 



,01 



1 . Low Involvement, No Decision, Positive 

2. High Involvement, No Decision, Positive 

3. Low Involvement, No Decision, Negative 

4. High Involvement, No Decision, Negative 

5. Low Involvement, Decision, Positive 

6. High Involvement, Decision, Positive 

7. Low Involvement, Decision, Negative 

8. High Involvement, Decision, Negative 



Discrepancy 1 .23 (0.25) [25." 

Discrepancy 1.05 (0.04) [25] 

Discrepancy 1 .14 (0.08) [22; 

Discrepancy 1 .29 (0.45) [25; 

Discrepancy 1 .29 (0.22) [25; 

Discrepancy 1.19 (0.26) [24; 

Discrepancy 1 .14 (0.12) [26; 

Discrepancy 1 .23 (0.36) [23; 



Main Effect Means 



Low Involvement 1.20 
High Involvement 1.19 



No Decision 
Decision 



1 .18 
1 .21 



Positive Discrepancy 1.19 
Negative Discrepancy 1.20 



Note: Analysis of mean responses to three Differential Emotions 
Scale items which were defined as comprising the disgust 
emotion: Feeling of distaste, feeling of revulsion, 
disgusted; Cronbach's alpha = .60. Significance of results 
is indicated by: a = p< . 0001 ; b = p< . 01 ; c = p<.05; d = 
p< . 1 . 



338 
of which returned a decision main effect (Tables 71, 72, 73). In each 
case, the mean response was higher for those conditions in which 
subjects made a decision than for those in which no decision was 
allowed, although none approached the previously discussed 3.0 level. 
Specifically, mean responses to the Joy items were 1.69 and 2.02; to 
Surprise 1.42 and 1.68; to Anger 1.06 and 1.20 for the no decision and 
decision conditions, respectively (variance explained ranged from 2 to 
3%). 

Mean responses to disgust revealed yet another pattern that was 
significant but accounted for a very small amount of variance (1%, 
Table 74). The effect that seemed to obtain was that under high 
involvement, the negative discrepancy conditions (i.e., the peanut 
coffee) created a significantly higher level of disgust (M = 1.26) than 
did the positive discrepancy conditions (M = 1.12). However, under low 
involvement, the reverse was true and disgust was higher in the 
positive discrepancy conditions (M = 1.26) than in the negative 
discrepancy conditions (M = 1.14), both pairwise differences are 
significant at a = .05 using Tukey's HSD test). A distraction 
explanation may well account for these results; but they are probably 
best considered trivial, given the very small proportion of variance in 
the measure that was explained. 

Overall, hypotheses 1a and 1b were not supported by these 
analyses. Only the decision manipulation demonstrated any consistent 
effect on the emotion scales as measured. Nearly identical results 
obtained from analyses of emotion scales responses normalized for each 
subject's mean emotional rating. Either the DES is not sensitive 



339 
enough for this task, or the manipulations were not strong enough to 
drive the process (or both). Implications for the conceptualization 
presented in the dissertation are best discussed in light of all 
results and are reserved for the general discussion and Chapter 7. 
Summary and Discussion 

Although the integrative interpretations are reserved for chapter 
7, some general conclusions are in order. The study suffered from some 
operational flaws resulting from the ambitious design. The effort to 
include three factors led to operationalization decisions (e.g., choice 
of target and distractor products) that may have weakened some of the 
tests. In particular, involvement is difficult to manipulate reliably 
and should probably be measured more accurately and used as a blocking 
variable or covariate. 

Overall, the hypotheses received mixed support. Hypotheses 1a and 
1b, predicting emotional patterns resulting from the manipulations, 
received little support from the DES measures--but the more common 
satisfaction scales indicated that some effect, at least, was present. 
The lack of such effects for the coffeemaker product validated further 
the contentions of hypotheses 1a and 1b, but the sensitivity of these 
measures was contrary to the prediction of hypothesis 2. Similarly, in 
evaluating hypothesis 3, greater effects on attitude measures than 
expected were found. 

The unexpectedly strong performance of the common satisfaction and 
attitude measures was repeated in correlation analyses with behaviors, 
hypotheses 4 and 5. Although emotion strengths did not correlate well 



340 
with behavior, these alternate measures did indicate some 
relationship. The analyses were hampered by low response rates. 

Among the supplemental hypotheses, the prediction of hypothesis 6 
that decision would enhance a contrast effect on product quality 
perception received no support. The contrast effect remains elusive 
and the decision opportunity manipulation had no apparent effect. The 
direct relationship between perception and attitude that was predicted 
by hypothesis 7 did receive strong support. Evidence for hypothesis 8, 
postulating a direct relationship between involvement and emotion 
strength, was weak. 

The measures of emotions used to detect effects on consumers' 
emotions did not show the responses that were predicted. There are at 
least two reasons for this: (1) the measures are not sufficiently 
sensitive to detect emotions evoked by consumption, and (2) no emotions 
were evoked. Both of these explanations reflect the fact that the 
measures were derived using a critical incident technique. It is, 
perhaps, true that the conditions evoked under those circumstances are 
more prominent than those felt in response to the typical consumer 
packaged good (although some descriptions did involve such products). 
In other words, true satisfaction or dissatisfaction may be relatively 
rare emotions and more difficult to arouse than so far argued. 

On a more positive note, the manipulation of discrepancy between 
expectations and perception seemed to mirror reasonably the results of 
prior investigations. While subjects were reluctant to admit to 
"error" in evaluating the poor quality product, their gift selections 
showed a marked trend away from the adulterated goods. The only 



341 
manipulation with a significant effect on gift selection was 
discrepancy. Eighty-four percent (84%) of the subjects in positive 
discrepancy conditions chose the test brand M (good quality coffee) as 
an honorarium for participation. In contrast, only 58% of the subjects 
in negative discrepancy conditions chose the brand that they had tested 
(poor quality — peanut coffee) as an honorarium (chi-square = 18.05, 1 
df, p<.0001; overall, chi-square = 22.66, 1 df, p< - 01 ) . 

Finally, the major contribution of the analyses, as a body, arises 
from the revealed effects of decision opportunity. While neither 
strong nor omnipresent, these effects were sufficient to demonstrate 
the importance of attending to decision as part of the consumption 
experience in designing a study. Including a purchase-like decision 
can help to eliminate a demand artifact (experimental situation 
effect) that need not exist in a consumer study. 

Chapter 7 evaluates these findings in light of the literature 
reviewed and the theory espoused. It concludes by offering proposals 
for improvements to the present design as well as directions for 
further investigation. 



CHAPTER 7 

CONCLUSION 

Introduction 

Based on an extensive review of the CS/D literature and related 
literatures, revisions to the dominant conceptualization of 
satisfaction (and dissatisfaction) were proposed. The essence of the 
new conceptualization was that the satisfaction/dissatisfaction 
continuum is the emotion-laden product of a specific psychological 
process. The intended contributions of the dissertation were the 
generation of measures of the construct and the delineation of elements 
necessary to the process. 

The research that was described in Chapter 5 and the results 
reported in Chapter 6 suggest additional modifications to the 
conceptualization. These results and their conceptual implications are 
discussed, improvements in the research design are proposed, and 
additional research that extends application of the conceptualization 
is suggested. 

Summary of Findings 
Pilot Study 

The original pilot study was conducted as a test of the critical 
incident technique and to identify the emotional profile of satisfied 
consumers and dissatisfied consumers. The critical incident 
methodology was successful at causing subjects to focus on memorable, 
emotion-laden events. Based on responses to the Differential Emotions 

342 



343 
Scale (DES), criterion levels of responses to emotion scales were 
established to define the endpoints of a satisfaction/dissatisfaction 
continuum. 

By its nature, the critical incident technique evokes memories of 
rare (critical) events. The motivation (involvement) level is thereby 
raised and emotions are likely to be strong relative to any aroused by 
less remarkable events. This fact should alert researchers to the 
difficulty of arousing satisfaction or dissatisfaction--indeed, any 
true emotion — in an experimental setting. This is not to say that such 
a manipulation is impossible to execute. Rather, the situation argues 
for either stronger manipulations that more reliably generate the 
desired effect or for more attention to the causes and effects of 
differences in responses by individuals. In addition, the study of 
satisfaction (especially, but not limited to CS/D) should probably take 
greater strides toward development of methodologies like critical 
incident which allow the individual respondent to reach higher arousal 
levels . 

As documented in the dissertation, the emotions evoked in the 
experimental setting were not strong. The strategies implemented to 
examine the effects of several factors and at the same time avoid 
experimental demand artifacts necessitated more subtle manipulations. 
The effects that obtained suggest that changes did occur in 
hypothesized, or at least understandable, patterns. This finding 
argues for a conclusion that broadens the dogmatic stand taken in 
Chapters 2 and 3. That is, rather than absolute thresholds defining at 
a universal level the presence or absence of satisfaction, the patterns 



344 
of emotions can be seen to vary through longer continua. Response to 
the strength of these emotions will vary by individual and by 
situation. Moreover, the strength of satisfaction (or dissatisfaction) 
reflects the influence of the manipulated factors. 
Manipulations 

Thus, although the measurement criteria are open to debate, the 
proposed conceptualization, "Satisfaction is the feeling which results 
from expectations that lead to a decision which is followed by personal 
experience in which the characteristics of the object of the decision 
are subject to perception which, in instances of sufficient response 
involvement allows for a conf irmat ion/ d is confirmation of the 
expectations and decision concerning the object," has not been 
rejected. Two of the three added constructs, decision and response 
involvement, have been found to affect some criterion measures of 
satisfaction . 

The strongest manipulation, determined by manipulation check 
measures as well as bv the effect on criterion variables, was decision 
opportunity. The two basic conditions, choice versus assignment of 
test product, accurately mirrored the methodology described for studies 
discussed in the literature review. Subjects were able to 
differentiate accurately between decision and non-decision conditions 
as well as identify the lack of decision opportunity regarding the 
cof feemaker . 

There was considerably more variation in the discrepancy 
manipulation. While the average results were directionally correct, 
individual-level data were not as encouraging. The manipulation of 



345 
discrepancy was able to account only for a small amount, less than 5%, 
of the variance in a scale designed to measure the perception of 
discrepancy between expectations and achieved performance. Measured 
expectations tracked well the intended manipulations. As determined in 
Chapter 6, the fluctuations centered on the perception of the coffee 
product rather than on expectations generated by advertisements and 
study instructions. This is likely to have resulted from an 
assimilation effect exaggerated by the limited personal experience with 
the product. This finding partially replicates the report of 
inaccurate product perception in Olson and Dover (1978). 

Note that while the results reflect limited experience with the 
product, the ability to personally perceive and evaluate the product 
was present. As chapter 3 argued, this is an antecedent to the 
satisfaction process. The perception of evidence as disconf irming may 
vary depending on the individual's relative confidence in the prior 
versus posterior information (Ofir 1982). The effect of personal 
experience versus second-hand information on the outcome of emotional 
processes is unknown. 

In order that such strong emotional processes may occur, 
sufficient motivation (involvement) must be present or aroused by the 
situation. The intended manipulation of involvement may not have 
occurred in the main study reported. The effect of the manipulation 
can be characterized as a variation in personal relevance that had the 
capacity to stimulate a limited variation in response involvement. The 
manipulation check measures adopted for this factor did not indicate 



346 
any reliable effect, although this may have been due to measure 
insensitivity . 

The problem that arises for satisfaction researchers is that there 
are no generally-accepted measures — nor, indeed, conceptualizations of 
involvement. The manipulation that was adopted was based on one by 
Petty et al. (1983) which was later associated with issue involvement, 
and not response involvement (Petty and Cacioppo 1986). Although the 
execution reflected the enduring involvement plus situation involvement 
definition derived by the dissertation, treatment by Petty and Cacioppo 
conflicts with this. Recent advances in measurement have focused en 
enduring rather than response involvement (e.g., McQuarrie and Munsor. 
1987, Zaichkowsky 1985, 1987). 

It may be to the advantage of CS/D researchers to focus on 
enduring involvement and its measures. Rather than seek to manipulate 
strong motivations and feelings, we might do better to measure them and 
attempt to explain them at the individual level. This argues for more 
attention to product, situation and individual interactions than has 
been paid previously. Individual differences in enduring involvement 
seem to account for more variance in response than manipulated 
situational involvement. The better strategy would be, perhaps, to 
utilize sampling plans resembling self-selected quasi-experimental 
designs (Cook and Campbell 1979) or make better use of critical 
incident techniques (Robinson 1976, 1980). The counterargument, of 
course, is that covariation, blocking, or other explicit recognition or 
enduring aspects of the subjects is a more tenuous experimental 
strategy than direct manipulation. 



347 
The problems experienced with the manipulation check measures 
recall the discussion presented cogently by Sternthal et al. (1987). 
Manipulation check measures can be interpreted as one indication of the 
status of the theoretical construct in the operationalization of an 
experiment (Perdue and Summers 1986). However, these are neither the 
only nor the most important indicants of the effectiveness of a 
manipulation. Measures in CS/D, as in consumer behavior in general, 
are not sufficiently well-developed that consensus exists regarding 
even the correct instruments to use. Thus, a comparative approach 
seems to be well in order. 
Tests of Hypotheses 

The main study sought to evaluate specific major and secondary 
hypotheses that had been derived from the revised conceptualization. 
As noted, these hypotheses garnered various levels of support resulting 
in mixed interpretations of the conceptualization. 

The first hypothesis, dealing with the criterion levels of the 
satisfaction emotions, found little evidence in its favor. It is 
possible to explain this in a number of ways: the conceptualization is 
simply wrong; problems with the DES indicate a lack of sufficient 
sensitivity to detect differences in consumer behavior; differences 
among individuals led to unreliability in manipulations and measures; 
and/or the situation did not arouse sufficiently strong emotions. Of 
these, the last is most likely and a potential underlying cause to two 
others. A weak manipulation of involvement could account for emotions 
too weak to be accurately measured or overwhelmed by individual traits. 



348 
The first explanation will not be entertained in its fullest at this 
time . 

What should be recognized and acknowledged at this juncture is the 
impact of involvement. The dissertation discussed the notion of a 
"conservation of motivation" as analogous to the laws of 
thermodynamics. Motivation may be stored and released but is neither 
created nor lost. The emotional expression of motivation represents an 
intermediate stage in the progression from involvement to behavior. 
Whereas the conceptualization, as presented, indicated a minimum 
threshold of involvement to enable the emotion-generating process 
(i.e., "sufficient" response involvement), a revised viewpoint would 
allow a more direct process. That is, strength of emotion: 1) varies 
with involvement (arousal); and 2) influences likelihood of behavior. 
From this standpoint, the negative findings on the first hypothesis can 
be interpreted as arousal of emotions too weak to be characterized by 
the individuals or determine their behavior (even such simple behaviors 
as completing the DES). 

The findings in tests of the second hypothesis bolster the 
foregoing contention. That is, the measures of satisfaction that were 
derived from the literature were not supposed to be able to detect 
variations in decision or response involvement. The fact that they did 
show a response to these manipulations, as well as to the discrepancy 
manipulation, is evidence that some consumer behavior process was 
affected. The lack of such influence on the coffeemaker satisfaction 
measures argues further for the process having been influenced by the 
experimental manipulations. 



349 
This pattern argues for modification of the stance taken by the 
dissertation. It had been argued that measures of attitude would not 
be distinguishable from the measures of satisfaction which were adapted 
from the CS/D literature. The emotional measures of satisfaction were 
expected to unconfound the constructs, operationalizing the conceptual 
distinction developed in Chapter 3. 

This dissertation posed and was attempting to help resolve the 
question of separability of the constructs. Similarity in behavior of 
the standard satisfaction measures and the attitude measures was 
expected for both the coffee and the coffeemaker; the emotional 
measures were expected to diverge from these. The sensitivity of both 
former sets of measures to the manipulation of decision opportunity in 
the case of coffee suggests great similarity in the processes 
influencing and influenced by each construct. This conclusion is 
consistent with Oliver's (1981) notion that satisfaction either decays 
into attitude or decays leaving attitude (analogous to the loss of 
energy through heat, friction, etc.). What is missing is evidence that 
the concepts are discrete at all. As emphasized from the introductory 
chapter forward, satisfaction is thought to comprise the more strongly 
motivational construct. Thus, there should be distinction in measures 
of emotional strength (the DES was not sensitive to this in this study) 
as well as in behavioral outcomes. 

Hypotheses 4 and 5 examined behaviors. As above, there were no 
distinct patterns that would differentiate satisfaction effects from 
attitude effects in this study. One pattern that did emerge simply 
confirmed that similarity in time and method (e.g., both written 



350 



responses) enhances correlation between attitude and behavior measures. 
Two satisfaction measures, Odds and D-T, showed slightly stronger 
correlations with behavior than did the other measures. This could be 
because they are anchored in more emotional terms and therefore more 
consistent with the conceptualization, or simply because there were 
more responses available to subjects and the improved correlations 
resulted from less restrictive measurement ranges. 

The hypothesized effect of decision on perceived discrepancy 
(hypothesis 6) simply did not occur; in fact the means suggested a 
slight assimilation rather than contrast effect. Because the decision 
opportunity manipulation was fairly strong elsewhere, failure to tmd 
the expected effect was not likely a result of the failure of the 
manipulation (see LaTour and Peat 1979 for a discussion of problems 
with contrast effects). This tentative hypothesis was identified as 
supplementary because of its exploratory nature. Thus, it is better 
to interpret the results as evidence against the particular process 
than as failure of either the manipulation or the measures. 

The measures used in evaluating hypothesis 7 were also questioned 
subsequent to the analysis. Some of the measures of perception could 
be construed as partially confounded with attitude due to the 
evaluative aspects of anchor points. These were the measures that 
correlated well with the defined attitude scales. Both products, the 
coffee and the coffeemaker, were chosen on the basis of their having 
relatively few features. This enabled quick, confident evaluation by 
the subjects and simpler manipulation of product quality in order to 
achieve the discrepancy manipulation. The cost of this strategy was 



351 

paid in not having sufficient perceptual features to provide a rigorous 
test of this supplemental hypothesis. Although the supporting 
correlations were high, the lack of correlation on some dimensions 
raises caution in accepting the simple hypothesis. 

Finally, the previously cited weakness in the measurement of 
response involvement can be implicated in the failure to find a 
correlation between involvement and emotion strength. In addition, the 
DES may share responsibility for lack of sensitivity in the present 
situation. If a more liberal interpretation of the conceptualization 
is accepted, then low response involvement may have generated low 
levels of emotions and both ranges were too restricted to show 
correlation. The harsher interpretation is that the involvement level 
was so low that no emotions were aroused. 

Implications of Findings 

The implications of the research can be summarized rather 
succinctly. The major contribution is the' identification of decision 
as an important factor in satisfaction processes. Most of the 
dependent measures were affected by this manipulation, either directly 
or through an interaction. Omission of decision in a research design 
that purports to evaluate satisfaction, and probably any post-purchase 
phenomenon, creates an experience different from what was intended. 

The importance of the decision construct in satisfaction research, 
or postpurchase research in general is highlighted in two very recent 
reports. Tse and Wilton (1988), in a study utilizing a role-playing 
task, report that product performance outperformed disconf irmation as a 
cause or predictor of satisfaction. This is not surprising, given the 



352 



research design that was reported. Subjects' expectations were 
manipulated by presentation of written product evaluations. Each 
subject was then assigned a product to evaluate, and allowed to examine 
and operate the item. As predictable by the conceptualization, 
performance perception—made concrete by personal experience—dominated 
the influence on satisfaction of abstract expectations or 
disconf irmation — unsecured by a decision. 

The importance of decision opportunity is echoed in a discussion 
by Oliver and DeSarbo (1988). They used decision opportunity, albeit 
in a role-playing context, to manipulate attribution of locus of 
control. Although the effect was weak, they concluded, "the locus of 
attribution effect displays the basic notion that persons are more 
satisfied when they feel responsible for the decision generating that 
satisfaction, ceteris paribus (p. 504)." The weakness of the effect 
that obtained may have been due, in part, to the use of a role-playing 
task. The lack of personal relevance likely led to low response 
involvement and, therefore, constrained satisfaction responses. 
Attempts to address further these issues relating to the effects of 
decision in the present research were largely unsuccessful for several 
reasons. The DES proved to be not sufficiently sensitive to detect 
low levels of emotions, if indeed these were present. The manipulation 
of response involvement through manipulation of personal relevance was 
not powerful enough to arouse emotions which would lead to reliable 
measures or predictable behaviors. 

Some success was achieved in implementing behavioral measures. 
While such behaviors as complaining or engaging in word-of-mouth are 



353 
denigrated in the CS/D literature as measures of satisfaction levels 
(due primarily to the restricted range which results from low incidence 
of such behaviors), the study reported here indicates that observation 
of behavior can be a useful measure when the perceived cost:benefit is 
approximately equal across individuals. 

An unexpected contribution of the dissertation arose from the 
analyses conducted in checking the discrepancy manipulation. After 
receiving the information in the advertisements and instructions, 
subjects claimed relatively simple, unidimensional expectations about 
both products, coffee and coffeemaker. After only a brief experience 
with each product, the subjects were prepared to make considerably more 
discriminating judgments. This finding has practical applications in 
both research and managerial settings, as it suggests storage or 
availability of more complex semantic information regarding the item. 

Research in Marketing frequently depends on single exposures to 
novel stimuli. The level of detail in the advertisements used in this 
study was not remarkably different from that in much of consumer 
research. Thus, it can be argued that the simple expectations that 
resulted were also typical of what might occur under laboratory 
conditions. The elaboration of the evaluative framework after a briet 
experience with the product suggests that efforts to compare 
expectations and perceptions directly may be misleading, at least when 
dealing with novel stimuli (Cadotte et al. 1983, Scott and Yalch 1930). 
For CS/D researchers interested in theory development, this is an 
interesting process to map and explain. For others, it is important to 
note that there is likely to be a considerably different evaluation 



354 



generated between a naive consumer and one with even minor personal 



experience . 



This difference has market implications as well. Advertisement 
pretesting frequently omits trial or disconf irmation as part of the 
design. Whereas an advertisement may elicit a generally positive 
response, failure or weakness on one or more attributes may create a 
decidedly more negative impression than once thought. Even if the 
post-purchase evaluation is compensatory, discriminable attribute 
scores allow for potentially unfavorable weightings. This argues for 
closer attention to consumer experience in the initial evaluation of 
products and communications. 

Future Outlook 
Revisions to Present Study 

If even a partial replication of the research reported here were 
to be undertaken, attention to and correction of some of the 
limitations would be necessary. The most crucial change would focus on 
response involvement. Alternatives to the strategy pursued in the 
present study, manipulating involvement within product, would involve 
manipulating involvement by using different products (although defining 
equivalence on all other dimensions would be nearly impossible) or by 
treating involvement as an individual difference variable in a blocking 
or covariate design. Any of these approaches requires an ability to 
measure the construct more accurately than is presently feasible. 
Measures of enduring involvement do not weigh the instantaneous arousal 
implied by the "conservation of motivation" formulation (e.g., Apsler 
and Sears 1968, Rhine and Severance 1970). 



155 



The DES needs additional development to be fully appropriate to 
marketing and consumer behavior situations. An expansion of the 
response categories may be adequate to increase the sensitivity of the 
instrument. Further development work could produce a research stream 
of its own. 

Similarly, the development of behavioral measurement technology 
for CS/D would be a fruitful line of inquiry to pursue. In the 
confines of improving the present study, however, it is sufficient to 
enhance the sensitivity of these measures as well. This can be 
accomplished by adopting a callback plan that generates a higher 
response rate, more strongly encouraging behaviors that do not require 
contact by the researcher (e.g., give each respondent a postcard that 
can be returned as a "vote" in favor of the product and one that votes 
against the product), or even better handling of unstructured 
responses . 



extensions 



Additional research beyond this basic design can also extend the 
findings or correct some of the limitations. The major finding was 
that decision opportunity affected many of the dependent variables of 
interest. The decision manipulation was strong, but simple. It would 
be of interest to examine the influence of variations of the decision 
opportunity along the lines proposed by Kiesler for the commitment 
construct (1971 — see Chapter 3). A more recent investigation, not yet 
fully analyzed nor documented, used the DES to investigate the 
satisfaction levels of recent homebuyers (recent was defined as within 
two years prior to the data collection). In this study, out of 162 



356 
respondents who met the ownership tenure criterion, 76 met or exceeded 
the 3.0 satisfaction criterion and only seven met the 3.0 
dissatisfaction criterion (interestingly, four met or exceeded both 
criteria) . Additional efforts to document use of the instrument in 
field settings would be helpful in improving the measures for general 
use. 

The influence of the apparent changes in cognitive structure 
should be examined more carefully. Why/how often does this occur? How 
soon does structure stabilize? Is this an example of schema formation? 
Does the satisfaction reaction generalize across the satisfaction types 
proposed in Table 2? On a similar note, is advertising perceived 
differently by novice versus experienced consumers? How much 
experience is enough to stabilize processing? 

Finally, what other factors are important in the satisfaction 
process? Do these models truly generalize across satisfaction domains 
(e.g., employee satisfaction, marital satisfaction, life satisfaction, 
etc.)? At what point do satisfaction and attitude merge? At what 
point do they separate? Are merge and separate adequate 
characterizations? 

In summary, there is still much to be learned and understood about 
satisfaction. What is needed is a fuller elaboration of a consensus 
paradigm that can provide ideas and strategies for research. 



APPENDIX A 
PILOT STUDY QUESTIONNAIRE 



Name (optional 

Sex 

Age 



CONSUMER RESEARCH PROJECT 



Please think about an occasion on which you, as a consumer, felt 
particularly dissatisfied with a product. Please answer the following 
questions with reference to this one occasion . 

1 . What type of product was involved? 



How long ago did this occur (approximately) 



Briefly describe the situation; be sure to mention anything you 
mav have done afterward. 



PLEASE DON'T GO ON UNTIL TOLD TO DO SO. 
357 



358 



Appendix A--continued 

This portion of the study consists of a number of words that 
describe different emotions or feelings. Please indicate the 
extent to which each word describes the way you felt on the 
occasion you described in question 3. 

Record your answers by circling the appropriate number on the 

five-point scale following each word. Presented below is the 

response scale for indicating the degree to which each word 
describes the way you felt. 

12 3 4 5 

Very Slightly Slightly Moderately Considerably Very 
or not at all Strongly 

In deciding on your answer to a given item or word, consider the 
feeling connoted or defined by that word. For example, if during 
the occasion you described in question 3 you felt very slightly or 
not at all "interested" you would circle the number 1 on the scale 
if you felt "interested" to a moderate degree you would circle 3; 
if you felt very strongly interested you would circle 5, and so 
forth. 

Remember, you should respond based on how you felt during the 
occasion you described in question 3. Please do not consider 
events that occurred afterwards, (e.g., a complaint that was 
resolved). Work at a good pace. It is not necessary to ponder; 
the first answer you decide on for a given word is probably the 
most valid. You should be able to finish this section in about 5 
minutes . 

Fiease turn to the next page and begin with the word marked by an 
arrow (-->). Choose a response for each word, in turn, until you 
reach the end of the list. 



359 



Appendix A--continued 



1 

Very Slightly 
or not at all 


2 
Slight 


iy 




Mod 


3 4 
erately Considerably 




5 
Very 
Strongly 


Interested 


2 


3 


4 


5 


Dissatisfied 


2 


3 


4 


5 


Contented 


2 


3 


4 


5 


Energetic 


2 


3 


4 


5 


Thrilled 1 


2 


3 


4 


5 


Discouraged 


2 


3 


4 


5 


Pleased 1 


2 


3 


4 


5 


Feeling of 
Aversion 


2 


3 


4 


5 


Joyful 1 


2 


3 


4 


5 


Sluggish 


2 


3 


4 


5 


Hostile 


2 


3 


4 


5 


Surprised 


2 


3 


4 


5 


Repentant 


? 


3 


4 


5 


Attentive 


2 


3 


4 


£ 


Angry 


? 


3 


4 


5 


Fearful 


2 


3 


4 


C 


Shy 


2 


3 


4 


5 


Contemptuous 


2 


3 


4 


5 


Haughty 


2 


3 


4 


5 


Downhearted 


2 


3 


4 


5 


B e f i a n t 


? 


3 


4 


5 


Scared 


2 


3 


4 


i 


Sheepish 


? 


3 


4 


5 


Afraid 


2 


3 


4 


5 


Blameworthy 


2 


3 


4 


5 


Feeling of 
Distaste 


2 


3 


4 


C 


Anxious 


2 


3 


4 


5 


Engaged in 
Thought 


2 


3 


4 


5 


Blissful 


2 


3 


4 


5 


Frightened 


2 


3 


4 


c; 


Rebellious 


2 


3 


4 


5 


Alert 


2 


3 


4 


5 


Guilty 


2 


3 


4 


5 


Sarcastic 


2 


3 


4 


5 


Wise 


1 2 


3 


4 


5 


Provoked 


2 


3 


4 


5 


Mad 


1 2 


3 


4 


5 


Foolish 


2 


3 


4 


5 


'"oncentrat ing 


1 2 


3 


4 


5 


Shocked 


1 2 


3 


4 


5 


Depressed 


1 2 


3 


4 


5 


Bashful 


1 2 


3 


A 


5 


Sad 


1 2 


3 


4 


5 


Inadequate 


1 2 


3 


4 


5 



360 



Appendix A — continued 



1 

Very Slightly 
or not at all 


Slig 


2 

htly 




3 

Moderately Cons 


4 
iderably 




5 
Very 

Strongly 


Mocking 


1 2 


3 


4 


5 


Scornful 


1 2 


3 


4 5 


Startled 


1 2 


3 


4 


5 


Annoyed 


1 2 


3 


4 5 


Bitter 


i 2 


3 


4 


5 


Fatigued 


1 2 


3 


4 5 


Irritated 


1 2 


3 


4 


5 


Disdainful 


1 2 


3 


4 5 


Sickened 


1 2 


3 


4 


5 


Astonished 


1 2 


3 


4 5 


Contemplative 


1 2 


3 


4 


5 


Emotional 


1 2 


3 


4 5 


Distressed 


1 2 


3 


4 


5 


Satisfied 


1 2 


3 


4 5 


Upset 


1 2 


3 


4 


5 


Happy 


1 2 


3 


4 


Shaky- 


i 2 


3 


4 


5 


Feeling of 
Revulsion 


1 2 


3 


4 S 


Disgusted 


1 2 


3 


4 


5 


Jittery 


1 2 


3 


4 5 


Excited 


1 2 


3 


4 


5 


Ashamed 


1 2 


3 


4 5 


Amazed 


1 2 


3 


4 


5 


Warmhearted 


1 2 


3 


4 5 


Lonely 


! 2 


3 


4 


5 


Feeling of 
Loathing 


1 2 


3 


4 5 


Enthusiastic 


1 2 


3 


4 


5 


Delighted 


1 2 


3 


4 5 


Enraged 


1 2 


3 


4 


5 


Awake 
Quarrelsome 


1 2 

1 2 


3 


4 5 
4 5 



Now, return to the first word in this section and continue 
circling responses for the remaining words. Only when you have 
circled a response for everv word on the page, go on to the rest 
of the questionnaire. 



Please respond to each of the following scales referring to the 
occasion you described in question 3. Again, please do not 
consider what may have occurred afterwards. Mark your response 
by checking the appropriate space or circling the appropriate 
number . 



36' 



Appendix A--continued 

a. How satisfied were you with the product? 

Very Somewhat Slightly Mixed Slightly 
Dissatisfied Dissatisfied Dissatisfied Satisfied 



Somewhat Very Neither Don't Know/ 
Satisfied Satisfied Don't Care 



i. Knowing what you know now if a similar situation arose, what ari 

the chances in ten (10) that you would purchase (use) the 
product again? 

0123456789 10 

No Chance Certain 



c. Overall, were you satisfied or dissatisfied with the product? 

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% .100% 

Completely Half & Completely 

Dissatisfied Half Satisfied 



Which word or phrase most closely reflects your satisfaction 
with the product? 



Delighted Pleased Mostly Mixed Mostly Unhappy Terrible 
Satisfied Dissatisfied 



Neutral Never Thought 
About It 



36: 



Appendix A--continued 

Please circle the word or phrase that most clearly reflects your 
agreement with each statement. 



I am satisfied with my decision to purchase (use) the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



2 If I had it to do all over again, I would feel differently about 
the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



3 My choice of the product was a wise one. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

- I feel bad about my decision concerning the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



5 I think that I did the right thing when I decided to get (use) 
the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



' I am not happy that I did what I did about the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

f. What would you normally do under the circumstances of the 

occasion you described in question 3? Please circle the word or 
phrase that most closely reflects your agreement with each 
statement no matter whether or no t you actually responded that 
way on this particular occasion . 

7 I would buy other products of the same brand. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

£ I would buy other products from the same seller. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



363 
Appendix A — continued 

9 I would buy other products of the same type. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

10 I would buy more of the same product as a gift. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

11 I would buy similar products as gifts. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

1 2 I would complain to a "higher authority" (owner, manager) 

Strongly Agree Agree Neither Disagree Strongly Disagree 

13 I would extend a compliment to a "higher authority" (owner, 
manager) . 

Strongly Agree Agree Neither Disagree Strongly Disagree 

14 I would tell a friend the good things about my experience. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

15 I would tell a friend the bad things about my experience. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

15 I would write directly to the manufacturer (original supplier) 
and complain. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

17 I would write directly to the manufacturer (original supplier) 
to compliment them. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



364 



Appendix A — continued 

- I would discard the product and replace it with a different 
brand as soon as possible. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



9 I would discard the product and replace it with a different 
product as soon as possible. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



2 3 I would recommend the product to a friend. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

21 I would recommend a friend avoid the product. 

Strongly Agree Agree Neither Disagree Strongly Disagree 

22 I would publicly endorse the product (e.g., in advertising). 

Strongly Agree Agree Neither Disagree Strongly Disagree 

23 I would publicly attack the product (e.g., in advertising). 

Stronglv Agree Agree Neither Disagree Strongly Disagree 

1L I would fight to have the product removed from market. 

Strongly Agree Agree Neither Disagree Strongly Disagree 



APPENDIX B 

INSTRUCTIONS FOR ADMINISTRATION OF 

EACH EXPERIMENTAL CONDITION 

A-H 



Set-Up Instructions: Cell 1 



Common Set-up: 



paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 



Special Set-up: 

gift packages: out of view 

drawing cards: visible 

coffee maker: red 

water bottle: red 

coffee instructions: red 

brand M: Uncle Charlie's 

brand N: generic 

ad booklet: white 



scissors 



365 



366 



Appendix B — continued 

SPECIFIC INSTRUCTIONS 

CELL 1 

CONDITION 001 

LOW INVOLVEMENT, NO DECISION, 

POSITIVE DISCREPANCY 



On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



The Subject completes the "Food Sensitivity Questionnaire". Study 
terminates if sensitivity to caffeine is indicated. 



The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our study. 
Today we are asking consumers like yourself to 
evaluate a coffeemaker that may be introduced by a 
major food processing company. The company has 
modified an existing design for easier use. In 
return for your assistance and taking the time to 
contribute your honest opinions , you will be given 
a chance to win a coffee maker in a drawing that 
will be conducted at the end of this phase of the 
study — that is, in about 4 weeks. 



367 



Appendix B — continued 



4. Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee 
maker and the company's own brand of coffee, also 
to be introduced soon. 

The coffee will be of adequate quality, but is 
primarily intended to be a cheap blend to help 
consumers cope with high prices brought about by 
the recent shortage. We are helping the company 
identify the minimal acceptable quality for the 
home market. 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads. 

Give the Subject the book of "negative" ads and the rating form. 



The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee, including 
brand M and brand N. Today we'd like for you to 
evaluate brand M. You've just read some 
advertisements for this brand. 

As you use the coffee maker, remember that you will 
have a chance to win one from the same company. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" form to 
comDlete . 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



Appendix B--continued 

8. The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

In addition to the chance at the coffee maker that 
you were promised, I'm happy to be able to offer 
you a week's supply of coffee as a gift. You may 
choose either brand M (which you've just tried) or 
brand N, a locally available generic brand. 

Mark the brand selected on the last page of the questionnaire at 
column 62. 



9. The subject is thanked for participation, entered into the coffee 
maker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product, 

Also, comment cards will be provided. Those returned within one 
week will provide a measure of complaining/ complimenting. 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



369 
Appendix B — continued 
Set-Up Instructions: Cell 2 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: visible 

drawing cards: out of sight 

coffee maker: red 

water bottle: red 

coffee instructions: red 

brand M: Uncle Charlie's 

brand N: generic 

ad booklet: white 

scissors 



370 
Appendix B--continued 



SPECIFIC INSTRUCTIONS 

CELL 2 

CONDITION 101 

HIGH INVOLVEMENT, NO DECISION, 

POSITIVE DISCREPANCY 



On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



2. The Subject completes the "Food Sensitivity Questionnaire' 
Study terminates if sensitivitv to caffeine is indicated. 



3. The cover story for the study is read to the Subject. 

Thank you for agreeing to participate in our study. 
Today we are asking consumers like yourself to 
evaluate coffee made by a major food processing 
company. The coffee will be of adequate quality, 
but is primarily intended to be a cheap blend to 
help consumers cope with high prices brought about 
by the recent shortage. We are helping them to 
identify the minimal acceptable quality for the 
home market. 

In return for your assistance and taking the time 
to contribute your honest opinions , you will be 
given a week's supply of the coffee. 



37' 



Appendix B — continued 



4. Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee and a 
related line of coffee makers. 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "negative" ads and the rating form. 



The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee, including 
brand M and brand N. Today we'd like for you to 
evaluate brand M. You've just read some 
advertisements for this brand. Remember, you will 
be given a week's supply of the brand you have 
evaluated. 

Give the Subject the gift bag at this point. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" form to 
complete. 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



372 
Appendix B — continued 

The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

At this time, I'd like for you to choose your gift 
of coffee. You may either keep brand M (which 
you've just tried) or choose brand N, a locally 
available generic brand. You will also receive a 
chance to win a coffee maker similar to this one in 
a drawing to be held at the end of this phase of 
the study, in about six weeks. 

Mark the selection on the last page of the questionnaire at 
column 62. 



9. The subject is thanked for participation, entered into the coffee 
maker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within om 
week will provide a measure of complaining/complimenting. 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



373 
Appendix B — continued 

Set-Up Instructions: Cell 3 



Common Set-up: 



paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: out of sight 

drawing cards: visible 

coffee maker: black 

water bottle: black 

coffee instructions: black 

brand M: peanut 

brand N: Food Club 

ad booklet: green 

can opener 



374 
Appendix B — continued 



SPECIFIC INSTRUCTIONS 

CELL 3 

CONDITION 002 

LOW INVOLVEMENT, NO DECISION, 

NEGATIVE DISCREPANCY 



1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



The Subject completes the "Food Sensitivity Questionnaire". 
Study terminates if sensitivity to caffeine or peanuts/peanut 
products is indicated. 



3. The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate a coffeemaker that may be 
introduced by a major food processing company. The 
company has modified an existing design for easier 
use. In return for your assistance and taking the 
time to contribute your honest opinions , you will 
be given a chance to win a coffee maker in a 
drawing that will be conducted at the end of this 
phase of the study — that is, in about 4 weeks. 



4. Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee 
maker and the company's own brand of coffee, also 
to be introduced soon. 



375 



Appendix B--continued 

The coffee will be very high in quality but only 
slightly more expensive than present top-of-the- 
line coffees. We are helping the company identify 
the quality most preferred by consumers. 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "positive" ads and the rating form. 



5. The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee. Today 
we'd like for you to evaluate brand M. You've just 
read some advertisements for this brand. 

As you use the coffee maker, remember that you will 
have a chance to win one from the same company. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" form to 
cormolete . 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

In addition to the chance at the coffeemaker that 
you were promised, I'm happy to be able to offer 
you a week's supply of coffee as a gift. You may 
choose either brand M (which you've just tried) or 
brand N, a popular brand in this area. 

Mark selection on the last page of the questionnaire at 
column 62. 



376 



Appendix B--continued 



9. The Subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts, and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting . 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



377 
Appendix B--continued 

Set-Up Instructions: Cell 4 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: visible 

drawing cards: out of sight 

coffeemaker: black 

water bottle: black 

coffee instructions: black 

brand M: peanut 

brand N: Food Club 

ad booklet: green 

can opener 



378 
Appendix B--continued 



SPECIFIC INSTRUCTIONS 

CELL 4 

CONDITION 102 

HIGH INVOLVEMENT, NO DECISION, 

NEGATIVE DISCREPANCY 



1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



The Subject completes the "Food Sensitivity Questionnaire. 
Study terminates if sensitivity to caffeine or peanut/ 
peanut products is indicated. 



3. The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate coffee made by a major food 
processing company. The company plans to introduce 
a very high quality coffee that is only slightly 
more expensive than present top-of-the-Iine 
coffees . 

We are helping them identify the quality most 
preferred by consumers. 

In return for your assistance and taking the time 
to contribute your honest opinions , you will be 
given a week's supply of the coffee. 

4. Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee and a 
related line of cof f eemakers . 



379 



Appendix B — continued 



On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "positive" ads and the rating 
form. 



5. The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee, including 
brand M and brand N. Today we'd like for you to 
evaluate brand M. You've just read some 
advertisements for this brand. Remember, you will 
be given a week's supply of the brand you have 
evaluated. 

Give the Subject the gift coffee at this point. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" 
form to comDlete. 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

At this time, I'd like for you to choose your gift 
of coffee. You may either keep brand M (which 
you've just tried) or choose brand N, a popular 
brand in this area. You will also receive a chance 
to win a coffeemaker similar to this one in a 
drawing to be held at the end of this phase of the 
study, in about six weeks. 

Mark the brand selected on the last page of the 
questionnaire at column 62. 



380 



Appendix B — continued 



9. The subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting. 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



381 
Appendix B--continued 

Set-Up Instructions: Cell 5 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: out of sight 

drawing cards: visible 

coffeemaker: red 

water bottle: red 

coffee instructions: red 

brand M: Uncle Charlie 

brand N: generic 

ad booklet: white 

scissors 



382 



Appendix B — continued 

SPECIFIC INSTRUCTIONS 

CELL 5 

CONDITION 01 1 

LOW INVOLVEMENT, DECISION, 

POSITIVE DISCREPANCY 

1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



The Subject completes the "Food Sensitivity Questionnaire. 
Study terminates if sensitivity to caffeine is indicated. 



3. The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate a coffeemaker made by a major 
food processing company. The company has modified 
an existing design for easier use. In return for 
your assistance and taking the time to contribute 
your honest opinions , you will be given a chance to 
win a coffeemaker in a drawing that will be 
conducted at the end of this phase of the study — 
that is, in about 4 weeks. 



4. Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffeemaker 
and the company's own brand of coffee, also to be 
introduced soon. 

The coffee will be of adequate quality, but is 
primarily intended to be a cheap blend to help 
consumers cope with high prices brought about by 
the recent shortage. We are helping the company 



383 



Appendix B — continued 

identify the minimal acceptable quality for the 
home market. 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads. 

Give the Subject the book of "negative" ads and the rating 
form. 



5. The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee, including 
brand M and brand N. M was the brand described in 
the ads which you've just read. N is a locally 
available generic coffee. We are, of course, most 
interested in your reaction to brand M, but since 
we're testing both M and N, you're free to choose 
either one. 

Please indicate the brand you have chosen in the 
space provided on the product evaluation form. 



Subject is given questionnaire and allowed to indicate 
choice of "Brand Selected" and "Brand Rejected". If Brand M 
is not selected, continue interview but do not count toward 
quota for this cell. 

As you use the coffeemaker, remember that you will 
have a chance to win one from the same company. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" 
form to complete. 

7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



384 
Appendix B — continued 

The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

In addition to the chance at the coffeemaker that 
you were promised, I'm happy to be able to offer 
you a week's supply of coffee as a gift. You may 
choose either brand M or brand N. 

Mark the brand selected on the last page of the 
questionnaire at column 62. 



9. The subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting . 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



385 
Appendix B--continued 

Set-Up Instructions: Cell 6 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: visible 

drawing cards: out of sight 

coffeemaker: red 

water bottle: red 

coffee instructions: red 

brand M: Uncle Charlie's 

brand N: generic 

ad booklet: white 

scissors 



386 
Appendix B — continued 



SPECIFIC INSTRUCTIONS 

CELL 6 

CONDITION 1 1 1 

HIGH INVOLVEMENT, DECISION, 

POSITIVE DISCREPANCY 



1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



2. The Subject completes the "Food Sensitivity Questionnaire. 
Study terminates if sensitivity to caffeine is indicated. 



3. The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate coffee made by a major food 
processing company. The coffee will be of adequate 
quality, but is primarily intended to be a cheap 
blend to help consumers cope with high prices 
brought about by the recent shortage. We are 
helping them to identify the minimal acceptable 
quality for the home market. 

In return for your assistance and taking the time 
to contribute your honest opinions , you will be 
given a week's supply of the coffee you choose. 



Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee and a 
related line of cof feemakers . 



387 



Appendix B — continued 

On the sheet provided, please write down what you 
think is the main-product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "negative" ads and the rating 
form. 



5. The Subject is shown the coffee choices. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee. Today we 
have both brand M and brand N. M was the brand 
described in the ads which you've just read. N is 
a locally available generic coffee. We are, of 
course, most interested in your reaction to brand 
M, but since we're testing both M and N, you're 
free to choose either one. 

Remember, you will be given a week's supply of the 
brand which you have chosen. Please indicate the 
brand you have chosen in the space provided on the 
product evaluation form. 

Subject is given questionnaire and allowed to indicate 
choice of "Brand Selected" and "Brand Rejected". If Brand M 
is not selected, continue interview but do not count toward 
quota for this ceil. 

Give the Subject the gift which was selected. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" 
form to complete. 



7. The Subject tastes the coffee which he/she has prepared. (The us 
of cream and/or sugar should be allowed, but not encouraged.) 



388 
Appendix B — continued 

8. The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

At this time, I'd like for you to confirm your 
choice of gift. You may either keep the brand you 
chose earlier or exchange it for the other brand. 
You will also receive a chance to win a coffeemaker 
similar to this one in a drawing to be held at the 
end of this phase of the study, in about six weeks. 

Mark the brand selected on the last page of the 
questionnaire at column 62. 

9. The subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 

10. After one week, delay measures are collected by telephone: 
self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting . 

11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



389 
Appendix B — continued 

Set-Up Instructions: Cell 7 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: out of sight 

drawing cards: visible 

coffeemaker: black 

water bottle: black 

coffee instructions: black 

brand M: peanut 

brand N: Food Club 

ad booklet: green 

can opener 



390 
Appendix B — continued 



SPECIFIC INSTRUCTIONS 

CELL 7 

CONDITION 012 

LOW INVOLVEMENT, DECISION, 

NEGATIVE DISCREPANCY 



1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



2. The Subject completes the "Food Sensitivity Questionnaire, 
Study terminates if sensitivity to caffeine or peanuts/ 
peanut products is indicated. 



The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate coffeemaker that may be 
introduced by a major food processing company. 
The company has modified an existing design for 
easier use. In return for your assistance and 
taking the time to contribute your honest opinions , 
you will be given a chance to win a coffeemaker in 
a drawing that will be conducted at the end of 
this phase of the study--that is, in about 4 
weeks. 



Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffeemaker 
and the company's own brand of coffee, also to be 
introduced soon. 



391 



Appendix B--continued 

The coffee will be very high in quality, but only 
slightly more expensive than present top-of-the- 
line coffees. We are helping the company identify 
the quality most preferred by consumers. 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "positive" ads and the rating 
form. 



5. The Subject is shown the coffee choices. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee. Today we 
have both brand M and brand N. M was the brand 
described in the ads which you've just read. N is 
a popular brand in this area. We are, of course, 
most interested in your reaction to brand M, but 
since we're testing both M and N, you're free to 
choose either one. 

Please indicate the brand you have chosen in the 
space provided on the product evaluation form. 

Subject is given questionnaire and allowed to indicate 
choice of "Brand Selected" and "Brand Rejected". If Brand M 
is not selected, continue interview but do not count toward 
quota for this cell. 

As you use the coffeemaker, remember that you will 
have a chance to win one from the same company. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" 
form to complete. 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



392 
Appendix B — continued 



The Subject is asked to complete the questionnaire containing the 
major measures. 

Offer the Subject a choice of gift coffee. 

In addition to the chance at the coffeemaker that 
you were promised, I'm happy to be able to offer 
you a week's supply of coffee as a gift. You may 
choose either brand M or brand N. 

Mark the brand selected on the last page of the 
questionnaire at column 62. 



9. The subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting. 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



393 
Appendix B — continued 

Set-Up Instructions: Cell 8 



Common Set-up: 

paper towels 

clean cup and spoon 

sugar, creamer, sweetener 

pencil 

coffee filters 

measuring spoons 

Special Set-up: 

gift packages: visible 

drawing cards: out of sight 

coffeemaker: black 

water bottle: black 

coffee instructions: black 

brand M: peanut 

brand N: Food Club 

ad booklet: green 

can opener 



394 

Appendix B — continued 

SPECIFIC INSTRUCTIONS 

CELL 8 

CONDITION 112 

HIGH INVOLVEMENT, DECISION, 

NEGATIVE DISCREPANCY 



1. On arrival, Subject is read informed consent: 

The study in which you are participating concerns 
your evaluations of certain common products. You 
will be asked to sample products and complete a 
questionnaire as part of the study. Your responses 
will be kept confidential. 

You are free to withdraw from the study at any 
time. There is no monetary compensation for 
participation. I will try to answer any questions 
you may have about the procedures. 



The Subject completes the "Food Sensitivity Questionnaire. 
Study terminates if sensitivity to caffeine or peanut/ 
peanut products is indicated. 



3. The cover story for the study is read to the subject. 

Thank you for agreeing to participate in our 
study. Today we are asking consumers like 
yourself to evaluate coffee made by a major food 
processing company. The company plans to introduce 
a very high quality coffee that is only slightly 
more expensive than present top-of-the-line 
coffees . 

We are helping them identify the quality most 
preferred by consumers. 

In return for your assistance and taking the time 
to contribute your honest opinions , you will be 
given a week's supply of the coffee you choose. 



Read the instruction, 

Before you actually try any product, we would like 
for you to look at some sample print 
advertisements. These ads are for the coffee and a 
related line of cof f eemakers . 



395 



Appendix B — continued 

On the sheet provided, please write down what you 
think is the main product feature that each ad is 
trying to communicate. Then rate how convincing 
you find the ad to be. Do this for each of the 
ads . 

Give the Subject the book of "positive" ads and the rating 
form. 



5. The Subject is shown the test coffee. 

Over the course of the study, we're comparing 
evaluations of several brands of coffee. Today we 
have both brand M and brand N. M was the brand 
described in the ads which you've just read. N is 
a popular brand in this area. We are, of course, 
most interested in our reaction to brand M, but 
since we're testing both M and N, you're free to 
choose either one. 

Remember, you will be given a week's supply of the 
brand which you have chosen. Please indicate the 
brand you have chosen in the space provided on the 
product evaluation form. 

Subject is given questionnaire and allowed to indicate 
choice of "Brand Selected" and "Brand Rejected". If Brand M 
is not selected, continue interview but do not count toward 
quota for this cell. 

Give the Subject the gift which was selected. 



6. Show Subject coffeemaker instructions and have Subject make a full 
(4-cup) pot. 

As the coffee brews, give the Subject the "Classification" 
form to complete. 



7. The Subject tastes the coffee which he/she has prepared. (The use 
of cream and/or sugar should be allowed, but not encouraged.) 



The Subject is asked to complete the questionnaire containing the 
maior measures. 



396 



Appendix B — continued 

Offer the Subject a choice of gift coffee. 

At this time, I'd like for you to confirm your 
choice of a gift. You may either keep the brand 
you chose earlier or exchange it for the other 
brand. You will also receive a chance to win a 
coffeemaker similar to this one at the end of this 
phase of the study, in about six weeks. 

Mark the brand selected on the last page of the 
questionnaire at column 62. 



9. The subject is thanked for participation, entered into the 
coffeemaker drawing, given gifts and dismissed. 

Interviewer initials the back of the questionnaire. 



10. After one week, delay measures are collected by telephone: 

self-report of Word-of-Mouth, 

self-report of the amount of use, 

self-report of purchase or attempt to purchase the product. 

Also, comment cards will be provided. Those returned within 
one week will provide a measure of complaining/ 
complimenting . 



11. After collection of the delay measures, the Subject will be 
questioned about suspicions regarding the hypotheses. The 
subject is then debriefed. 



APPENDIX C 
QUESTIONNAIRE AND SUBJECT FORMS FROM MAIN STUDY 

University of Akron 

Division of Consumer Testing 

Food Sensitivity Questionnaire 

Many of the product studies which we administer include an 
opportunity to sample various foods. When this is the case, we ask 
participants to notify us of any known allergies or other dietary 
restrictions. You will not be asked to try any products containing 
those ingredients you have told us about. 

PLEASE PLACE A CHECK (vO NEXT TO ANY OF THE FOLLOWING 
WHICH YOU CANNOT OR PREFER NOT TO EAT FOR ANY REASON . 
SIMPLY MARK EACH SUCH ITEM. YOU WILL NOT BE ASKED TO 
EXPLAIN ANY SELECTION. 



Sugar 
' Salt 
Caffeine 
Chocolate 
Saccharin 
Aspartame 
(NutraSweet) 
MSG (monosodium 

glutamate) 
Other artificial 

sweeteners 
Other artificial 
preservatives 



Fish 
Beef /beef 

derivatives 
Pork/pork 

derivatives 
Poultry/poultry 

derivatives 
Liver/other 

organ meats 
Milk/dairy 

products 
Calcium 
Eggs/egg 

derivatives 



Peanuts/peanut 
products 
Other nuts. 'nut 

products 
Vegetat les 

( specify) 

. Fruits (specify) 

Soybean/ Soybean 
derivatives 



Please note here any known dietary restriction not covered above. 



Signature 



Date 



397 



393 



Appendix C — continued 
Advertisement Rating Form 

Please evaluate each advertisement individually 
Ad #1 

What product feature(s) is this ad trying to communicated 



Place a check (•") in the space that indicates how convincing you find 

this advertisement to be. 

Not at all Somewhat Moderately Extremely 
Convincing Convincing Convincing Convincing 



Ad #2 

What produce feature(s) is this ad trying to communicate? 



Place a check (•) in the space that indicates how convincing you find 

this advertisement to be. 

Not at all Somewhat Moderately Extremelv 
Convincing Convincing Convincing Convincing 



Ad #3 

What product feature(s) is this ad trying to communicate: 



Place a check (•) in the space that indicates how convincing you find 
this advertisement to be. 

Not at all Somewhat Moderately Extremely 

Convincing Convincing Convincing Convincing 



Ad #4 

What product feature(s) is this ad trying to communicate? 



Place a check (vO in the space that indicates how convincing you find 

this advertisement to be. 

Not at all Somewhat Moderately Extremely 
Convincing Convincing Convincing Convincing 



399 



Appendix C — continued 
Classification Information 



Name : 

Phone : 

Mailing address: 

Sex: M F 

Age: 



Employed: Yes No 



Average number of cups of coffee (all kinds) you drink each day: 



Do you prepare coffee yourself: 
at home? 

always 

sometimes 

never 

at work? 

always 

sometimes 

never 

What brand(s) of coffeemaker do you/have you used? 



What brand(s) of coffee do you/have you used' 



Please list every brand of coffee you can think of: 



Please list every brand of coffeemaker vou can think of: 



Please check the spaces that describe what you would consider to be an 
"ideal" coffee. 

For me, an ideal coffee is: 

Moder- Moder- 

Very ately Slightly Neither Slightly ately Very 

Bitter _ Not 

Bitter 

Strong in Weak 

in flavor " in flavor 

Poor _ Good 

Quality ~~ " Quality 

Light in _ Dark in 

Color ~" Color 

Good _ _ _ Bad 

Tasting ' ' " Tasting 



400 

Appendix C — continued 

BRAND SELECTED 

BRAND REJECTED 

Questionnaire 

Thank you for agreeing to participate in our market test. At 
this point, we would like to know your reactions to the product you 
have tried and to other aspects of the market test. 

Part I 
Directions : 

In this section, please place a check (/) in the space that best 
represents your opinion or reaction to parts of the study. For 
example, if you feel that today's weather is Very Sunny but Moderately 
Cold , you would mark the appropriate scales as follows: 

Today's weather is 

Very Moderately Slightly Neither Slightly Moderately Very 

Cloudy _ \/_ Sunny 

Hot _v/ Cold 

Please mark only one space on each line, but be certain not to 
miss any line. Remember, base your answers on today's study. It you 
have any questions, ask the interviewer: 

The Coffeemaker is: 

Moder- Moder- 

Very ately Slightly Neither Slightly atelv Very 

Difficult _' Simple to 

to Use L 'se 

Fast _ s l° w 

Simple to _ ~ ~~ "_ Difficult 

to Fill '" ' to Flll_ 

Unattractive _ Attractive 

Better __ ~~ "_ _ Worse than 

than I expected ~" I expected 

The Coffee is: 

Moder- Moder- 

Very ately Slightly Neither Slightly ately Very 

Bitter _ Not Bitter 

Strong in " ~ _ Weak in 

Flavor "~ Flavor 

Poor Good 

Quality ~ " Quality 

Light in Dark in 

Color ~" Color 

Good Bad 

Tasting " Tasting 

Worse than Worse than 

I expected . . . - I expected 



401 
Appendix C--continued 
COFFEE BRAND EVALUATED 



Questionnaire 

Thank you for agreeing to participate in our market test. At this point, 
we would like to know your reactions to the product you have tried and to other 
aspects of the market test. 

Part I 
Directions 

In this section, please place a check (•/ ) in the space that best 
represents your opinion or reaction to parts of the study. For example, if you 
feel that today's weather is Very Sunny but Moderately Cold , you would mark the 
appropriate scales as follows: 

Today's weather is 

Very Moderately Slightly Neither Slightly Moderately Very 
Cloudy <y Sunny 

Hot ""_"_ _"" '""" ~~ _ k / ~"_ Cold' 

Please mark only one space on each line, but be certain not to miss any 
line. Remember, base your answers on today's study. If you have any 
questions, ask the interviewer: 

The Coffeemaker is: 

Moder- Moder- 
Very ately Slightly Neither Slightly ately Very 

Difficult _ Simple to 

to Use Use 

Fast _ Slow 

Simple to _ Difficult 

to Fill ~~ "~ ~' ~~ to Fill 

Unattractive _ Attractive 

Better _ Worse than 

than I expected I expected 

The Coffee is: 
Moder- Moder- 
Very ately Slightly Neither Slightly ately Very- 
Bitter _ _ Not Bitter 
Strong in _ Weak in 
Flavor Flavor 
Poor _ Good 
Quality Quality 
Light in _ _ Dark in 
Color Color 
Good _ _ Bad 
Tasting Tasting 
Worse than _ Worse than 
I expected I expected 



402 



Appendix C--continued 
Part II 

Directions 

This portion of the study consists of a number of words that 
describe emotions or feelings. Please indicate the extent to which 
each word describes the way you feel about the coffee that you have now 
sampled. 

Record your answer by circling the appropriate number on the five 
point scale following each word. Presented below is the response 
scale for indicating the degree to which each word describes the way 
you feel about the coffee. 

12 3 4 5 

Not at all or Very- 

Very Slightly Slightly Moderately Considerably Strongly 

In deciding on your answer to a given item or word consider the 
feeling connoted or defined by that word. Then, if sampling the 
coffee makes you feel that way not at all or very slightly you would 
circle the number 1 on the scale; if you feel that way to a moderate 
degree, you would circle 3; if you feel that way very strongly you 
would circle 5, and so forth. 

Remember, you are requested to make your responses on the basis of 
the way you feel about the coffee you sampled. Work at a good pace. 
It is not necessary to ponder; the first answer you decide on for a 
given word is probably the most valid. You should be able to finish 
this section in about 3 minutes. 

Sampling the coffee makes me feel: 



Not at all or 










Very 


Very Slightly 


Slightly 


Moderately 


Consider 


ably 


Strongly 


Discouraged 1 


2 


3 


4 




5" 


Joyful 1 


2 


3 


4 




5 


Enraged 1 


2 


3 


4 




5 


Thrilled 1 


2 


3 


4 




5 


Contented 1 


2 


3 


4 




5 


Interested 1 


2 


3 


4 




c 


Guilty 1 


2 


3 


4 




5 


Downhearted 1 


2 


3 


4 




5 


Sad 1 


2 


3 


4 




5 


Satisfied 1 


2 


3 


4 




5 


Attentive 1 


2 


3 


4 




5 


Repentant 1 


2 


3 


4 




5 


Scared 1 


2 


3 


4 




5 


Delighted 1 


2 


3 


4 




5 


Disgusted 1 


2 


3 


4 




5 


Pleased 1 


2 


3 


4 




5 


Angry 1 


2 


3 


4 




5 


Disdainful 1 


2 


3 


4 




5 



403 



Appendix E — continued 



Not at all or 






Very Slightly 


Slightly 


Moderately 


Alert 1 


2 


3 


Absorbed 1 


2 


3 


Happy 1 


2 


3 


Sheepish 1 


2 


3 


Fearful 1 


2 


3 


Blameworthy 1 


2 


3 


Bashful 1 


2 


3 


Amazed 1 


2 


3 


Concentrating 1 


2 


3 


A Feeling of 






Distaste 1 


2 


3 


Surprised 1 


2 


3 


Dissatisfied 1 


2 


3 


Scornful 1 


2 


3 


Involved 1 


2 


3 


Afraid 1 


2 


3 


A Feeling of 






Revulsion 1 


2 


3 


Shy 1 


2 


3 


Contemptuous 1 


2 


3 


Astonished 1 


2 


3 


Mad 1 


2 


3 



Very 

Considerably Strongly 
4 5 

4 5 

4 5 

4 5 

4 5 

4 5 

4 5 

4 5 

4 5 



Part III 

Directions 

Please respond to each of the following scales based on the 

products you tested today. Mark your response by placing a check (vO 

in the appropriate space or by circling the appropriate number. 



How satisfied were you with the coffee 



Very 
Dissatisfied 



Somewhat 
Dissatisfied 



Slightly 
Dissatisfied 



Mixed 



Slightly- 
Satisfied 



Somewhat 
Satisfied 



very 
Satisfied 



Not Satisfied 
Nor Dissatisfied 



404 



Appendix C — continued 

How satisfied were you with the coffeemaker? 

Very Somewhat Slightly Slightly- 

Dissatisfied Dissatisfied Dissatisfied Mixed Satisfied 



Somewhat Very Not Satisfied 
Satisfied Satisfied Nor Dissatisfied 



Knowing what you know now, what are the chances in ten (10) that you 
would choose to use the coffee again? 

0123456789 10 

No Certain 

Chance 



Knowing what you know not, what are the chances in ten (10) you would 
choose the coffeemaker again? 

1 2 3 4 5 6 7 8 9 10_ 

n; Certain 

Chance 

Which word or phrase most closely reflects your satisfaction with the 
coffee? 

Mostly Mostly 

Delighted Pleased Satisfied Mixed Dissatisfied Unhappy Terrible 

Neutral 

Never 

Thought 

About It 



Which word or phrase most closely reflects your satisfaction with the 
coffeemaker? 

Mostly Mostly 

Delighted Pleased Satisfied Mixed Dissatisfied Unhappy Terrible 

Neutral 

Never 

Thought 

About It 



405 



Appendix C — continued 
Which face comes closest to expressing how you feel about the coffee? 






Which face comes closest to expressing how you feel about the 
cof feemaker? 







Part IV 

Directions 

This section consists of words and phrases that describe the study 
and the products. Please place a check (• ) on the portion of the 
scale that best reflects your feelings. 

Evaluation of the Coffee 

Was interesting to me Was not interesting 

to me 

Was important to me Was not important to 

me 

Was not difficult to Was difficult to 

evaluate " evaluate 

Took a lot of Did not take a lot 

thought of thought 

Task was very ^ Task was not very 

involving ~ involving 

I put a lot of I P ut no thought 

thought into into evaluation 
evaluation 



406 



Appendix C — continued 

Evaluation of the Coffeemaker 

Was interesting to me _ Was not interesting 

to me 

Was important to me Was not important to 

me 

Was not difficult to Was difficult to 

evaluate " evaluate 

Took a lot of Did not take a lot 

thought "~ of thought 

Task was very _ Task was not very 

involving involving 

I put a lot of _ I put no thought 

thought into into evaluation 
evaluation 



Coffee 

Good _ Bad 

Unsatisfactory _ """ Satisfactory 

Unfavorable _ Favorable 

Worthless ~ "" Valuable 

Like _ Dislike 

Disapprove '_ Approve 



Coffeemaker 

Good Bad 

Unsatisfactory Satisfactory 

Unfavorable ~~ "" " Favorable 

Worthless Valuable 

Like Dislike 

Disapprove " Approve 



407 

Appendix C — continued 

Part V 

Directions 

The questions in this section refer to what you expected before 
trying the coffeemaker and coffee. Please place a check (>/) on the 
portion of each scale that best reflects your feelings. 

Based on the information given to you in the advertisements at the 
beginning of the study, what did you expect about the coffeemaker? Did 
you expect it to be 

Moder- Moder- 

Very ately Slightly Neither Slightly ately Very 

Difficult _ Simple to 

to Use Use 

Fast _ slow 

SimDle to . Difficult 



to Fill 



to Fill 



Unattractive __ Attractive 

Better Worse than 

than I "~ : expected 

expected 

The Coffee is: 
Moder- Moder- 

Very ately Slightly Neither Slightly ately Very 

Bitter 1 Not Bitter 

Strong in "_ ~~ Weak in 

Flavor """ Flavor 



Poor 



lood 



Quality " " Quality 

Light in _ Ddrk in 

Color " ~ Color 



Good 



Bad 



Tasting Tasting 

Inexpensive _ _ expensive 

Overall, my expectations about the coffeemaker were: 

Too high; To ° lov; '> 

It was worse Accurate It was better 

Than I thought Than l thought 



Overall, my expectations about the coffee were: 

Too high; Too low; 

It was worse Accurate It was better 

Than I thought Tha " x thought 



408 



Appendix C--continued 

Part VI 

Directions 

In this last section, please circle the word or phrase that most 
closely reflects your agreement with each statement. 

The next time I need coffee, I will consider buying the brand I tried 
(if available) . 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

The next time I need a coffeemaker for myself, I will consider buying the 
model I tried (if available). 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

The next time I serve coffee to guests, I would use the brand I tried. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

If I wanted to give a coffeemaker as a gift, I would consider giving the 
model I tried. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 



Overall, my experience in the study was enjoyable. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 



I had the option to leave the study at any time. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 



I had the option to choose the brand of coffee to evaluate. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 



409 

Appendix E — continued 

I had to option to choose the model of coffeemaker to evaluate. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

Given the opportunity, I would participate in another product study. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

I am satisfied about my choice of coffee. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

If I had it to do all over again, I would feel differently about choosing 
the coffee. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

My choice of coffee was a wise one. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

I feel bad about my decision regarding the coffee. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

I think that I did the right thing when I chose the coffee. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 

I am not happy that I chose the coffee that I did. 

Strongly Moderately Slightly Slightly Moderately Strongly 

Agree Agree Agree Neither Disagree Disagree Disagree 



410 



Appendix C — continued 



Sirs : 

I recently had the opportunity to try 
your new coffee. I wanted to make the 
following comments about it: 



Market 


Research 














(st 


amp) 


University 


of Ak 


ron 
















Akron, 


OH 


44325- 


1625 






















NEH 






















P.O. 


Box 5 


4 


















Fern 


Park, 


FL 


32' 


'30 







41 1 



Appendix C — continued 
Follow-Up Interview 



Subject number: 

Date of first interview: 

Name : 

Phone : 



Call back date: 



1. time: result: 

2. time: result: 

3. time: result: 

4. time: result: 

5. time: result: 

Get the participant on the line, then: 

Good (morning, afternoon, evening), I'm calling for NEH research. 
One week ago, you participated in a study which we are conducting. I'd 
like to ask you 6 brief questions about the coffee you tried at that 
time . 

1 . How many times have you told someone about the coffee you tried a 
week ago? (number of people) 

2. What did you tell them? circle: positive negative 

3. How manv times have you used the gift sample you were given (# 
pots): 

4. Have you served it to: others in household? yes no 
guests? yes no 

5. If you had the opportunity, would you consider repurchasing the 
coffee? yes no 

6. Have you tried to repurchase the coffee? yes no 

At this time, I want to inform you that this is not a test market 
study being conducted by a food processing company. Rather, this is 
part of an academic study of customer satisfaction being conducted at 
the University of Akron. The advertisements and products which you 
evaluated were prepared for this study. The coffee sample you were 
given in the paper bag was not modified and may be used per the label 
instructions. A drawing for a coffeemaker will be held as soon as the 
study is completed. If you have any questions, I can provide you with 
the name and phone number of the professor conducting the study. If 
ycu have not yet mailed the postcard that you were given, please 
discard it. We are only interested in your response prior to this 
call. Thank you again for your assistance. (Doug Hausknecht 375-6303) 



412 



Appendix C — continued 



Marketing Department 
University of Akron 
Akron, OH 44325 



(stamp) 



(participant's address) 



Dear 

We have attempted to contact you by telephone in 
regard to the coff ee/cof feemaker study in which you 
participated. In addition to a few follow-up questions, 
we wished to inform you that the study was conducted as 
part of academic research on consumer satisfaction/ 
dissatisfaction and was not at the request of any 
manufacturer. If you have not yet returned your 
"comment card", please discard it at this time. 

The coffee you received as a gift (foil pouch in the 
paper bag) was not altered and may be used as indicated 
in the directions. A drawing for a coffeemaker will be 
held when all the data are collected. If you have any 
questions about the study, please call Professor 
Hausknecht at the University of Akron (216) 375-6303. 

Thank you for vour assistance. 



REFERENCES 

Adler, Roy D. and Larry M. Robinson (1980), "A Consumer Satisfaction 

Model Based on Job Satisfaction Theory," in Refining Concepts and 
Measures of Consumer Satisfaction and Complaining Behavior , eds. 
H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana 
University, 19-22. 

Aiello, Albert Jr. and John A. Czepiel (1979), "Customer Satisfaction 
in a Catalog Type Retail Outlet: Exploring the Effect of Product, 
Price and Attributes," in New Dimensions of Consumer Satisfaction 
and Complaining Behavior , eds. Ralph L. Day and H. Keith Hunt, 
Bloomington, IN: Indiana University, 129-135. 

John A. Czepiel, and Larry J. Rosenberg (1977), "Scaling 



the Heights of Consumer Satisfaction: An Evaluation of Alternative 
Measures," in Consumer Satisfaction, Dissatisfaction and 
Complaining Behavior , ed. Ralph L. Day, Bloomington, IN: Indiana 
University, 43-50. 

Alba, Joseph W. and J. Wesley Hutchinson (1987), "Dimensions of 

Consumer Expertise," Journal of Consumer Research 13 (March), 411- 
454. 

Alderson, Wroe (1965), Dynamic Marketing Behavior: A Functionalist 
Theory of Marketing , Homewood, IL: Richard D. Irwin, Inc. 

Allen, Chris T., Karen A. Machleit, and Susan S. Marine (1987), "On 

Assessing the Emotionality of Advertising Via Izard's Differential 
Emotions Scale," paper presented at Association for Consumer 
Research Conference, Cambridge, MA: Oct. 8-11. 

Anderson, Rolph E. (1973), "Consumer Dissatisfaction: The Effect of 

Disconfirmed Expectancy on Perceived Product Performance," Journal 
of Marketing Research 10 (February), 38-44. 

, and Joseph F. Hair, Jr. (1972), "Consumerism, Consumer 

Expectations and Perceived Product Performance," in Association 
for Consumer Research, Annual Proceedings #3 , ed. M. Venkatesan, 
Ann Arbor, MI: Association for Consumer Research, 67-79. 

Anderson, Ronald D., Jack L. Engledow and Helmut Becker (1979), 

"Evaluating the Relationships Among Attitude Toward Business, 
Product Satisfaction, Experience and Search Effort," Journal of 
Marketing Research 16 (August), 394-400. 



413 



414 



Andreasen, Alan R. (1977a), "A Taxonomy of Consumer Satisfaction/ 
Dissatisfaction Measures," Journal of Consumer Affairs 11 
(Winter), 11-24. 

, (1977b), "A Taxonomy of Consumer Satisfaction/ 



Dissatisfaction Measures," in Conceptualization and Measurement of 
Consumer Satisfaction and Dissatisfaction , ed. H. Keith Hunt, 
Cambridge, MA: Marketing Science Institute, 11-35. 

, and Arthur Best (1977), "Consumers Complain - Does 
Business Respond?" Harvard Business Review 55 (July-August), 
93-101 . 

Andrews, Frank M. and Stephen B. Witney (1976), Social Indicators of 
Well-Being , New York, NY: Plenum Press. 

Apsler, Robert and David 0. Sears (1968), "Warning, Personal 

Involvement, and Attitude Change," Journal of Personality and 
Social Psychology 9 (Number 2), 16 2-166. 

Arndt, Johan (1967), "Role of Product-Related Conversations in the 
Diffusion of a New Product," Journal of Marketing Research 4 
(August), 291-295. 

Aronson, Elliot (1968a), "Discussion: Expectancy vs. Other Motives," 
in Theories of Cognitive Consistency: A Sourcebook , eds. Robert 
P. Abelson et al., Chicago, IL: Rand McNally and Company, 491-493. 

and J. Merrill Carlsmith (1962), "Performance Expectancy 

as a Determinant of Actual Performance," Journal of Abnormal and 
Social Psychology 65 (Number 3), 178-182. 

Assar, Amardeep and Dipankar Chakravarti (1 984), "Attribute Range 

Knowledge: Effects on Consumers' Evaluation of Brand-Attribute 
Information and Search Patterns in Choice," in American Marketing 
Association 1984 Educators' Conference Proceedings , ed. Russell W. 
Belk, Chicago, IL: American Marketing Association, 62-67. 

Atkinson, John W. (1957), "Motivational Determinants of Risk-Taking 

Behavior," reprinted in Research in Consumer Behavior , eds. David 
T. Kollat, Roger D. Blackwell, and James F. Engel, New York, N.Y.: 
Holt, Rinehart and Winston, Inc., 31-47. 

, and David Birch (1978), An Introduction to Motivation 2nd 



edition, N<;w York. NY: Van Nostrand. 

Averill, James R. (1980), "A Constructivist View of Emotion," in 
Emotion: Theory, Research, and Experience Volume 1, eds. 
Robert Plutchik and Henry Ke Herman, New York, NY: Academic 
Press, 305-339. 



415 



Bahr, William J. (1982), "Consumer Dissatisfaction or Disappointment: 

The Critical Difference," in New Findings on Consumer Satisfaction 
and Complaining , eds . Ralph L. Day and H. Keith Hunt, Bloomington, 
IN: Indiana University, 33-39. 

Bar-Hillel, Maya (1980), "The Base-Rate Fallacy in Probability 
Judgments," Acta Psychologica 44, 211-233. 

3arbeau, J. Bradley and William J. Quails (1984), "Consumers' 

Perceptions of Attributes and Behavioral Intentions: An Extended 
Comparison-Level Model," in Advances in Consumer Research: Volume 
1_1_, ed. Thomas C. Kinnear, Provo, UT: Association for Consumer 
Research, 143-147. 

Barnes, James G. and Karen R. Kelloway (1980), "Consumerists : 

Complaining Behavior and Attitudes Toward Social and Consumer 
Issues," in Advances in Consumer Research: Volume 7 , ed. Jerry C. 
Olson, Ann Arbor, MI: Association for Consumer Research, 329-334. 

Bearden, William 0. (1983), "Profiling Consumers Who Register 

Complaints Against Auto Repair Services," Journal of Consumer 
Affairs 17 (Winter), 315-335. 

, and J. Barry Mason (1984), "An Investigation of 

Influences on Consumer Complaint Reports," Advances in Consumer ^ 
Research: Volume 1 1 , ed. Thomas C. Kinnear, Provo, UT: Association 
for Consumer Research, 490-495. 

and Richard L. Oliver (1985), "The Role of Public and 

Private Complaining in Satisfaction with Problem Resolution," 
Journal of Consumer Affairs 19 (Winter), 222-240. 

, and Jesse E. Teel (1983), "Selected Determinants of 

Consumer Satisfaction and Complaint Reports," Journal of Marketing 
Research 20 (February), 21-28. 

Becherer, Richard C, Fred W. Morgan, and Lawrence M. Richard (1982), 
"The Job Characteristics of Industrial Salespersons: Relationship 
to Motivation and Satisfaction," Journal of Marketing 46 (Fall), 
125-135. 

Belk, Russell W. (1984a), "Three Scales to Measure Constructs Related 
to Materialism: Reliability, Validity and Relationships to 
Measures of Happiness," in Advances in Consumer Research: Volume 
1_1_, ed. Thomas C. Kinnear, Provo, UT: Association for Consumer 
Research, 291-297. 

(1984b), "Manifesto for a Consumer Behavior of Consumer 

Behavior," in Proceedings of 1984 Winter Educators' Conference: 
Scientific Method in Marketing , eds. Paul F. Anderson and Michael 
J. Ryan, Chicago, IL: .American Marketing Association, 163-167. 



416 



(1984c), "Against Thinking," in Proceedings of 1984 

Winter Educators' Conference: Scientific Method in Marketing , 
eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American 
Marketing Association, 57-60. 

(1986), "Art Versus Science as Ways of Generating 
Knowledge About Materialism," in Perspectives on Methodology in 
Consumer Research , eds. David Brinberg and Richard J. Lutz, New 
York, NY: Springer-Verlag, 3-36. 

, John F. Sherry, Jr. and Melanie Wallendorf (1988), "A 

Naturalistic Inquiry into Buyer and Seller Behavior at a Swap 
Meet," Journal of Consumer Research 14 (March), 449-470. 

, Melanie Wallendorf, John Sherry, Morris Holbrook, and 
Scott Roberts (1988), "Collectors and Collecting," in Advances in 
Consumer Research , Volume 15, ed. Michael J. Houston, Provo, UT: 
Association for Consumer Research, 548-553. 

Bergh, Bruce G. Vanden and Leonard N. Reid (1980), "The Roleof 

Newspaper Advertisements in the Confirmation or Disconf irmation of 
Consumer Expectations in a Product Trial Situation," in 
Advertising 1980: Voice of a Nation at Work , ed. James E. Haefner, 
Urbana, IL: University of Illinois, 61-64. 

Bernacchi, M. D., Ken Kono, and G. L. Willette (1980), "An Analysis of 
Automobile Warranty Service Dissatisfaction," in Refining Concepts 
and Measures of Consumer Satisfaction and Complaining Behavior , 
eds. H. Keith Hunt and Ralph L. Day, Bloomington, IX: Indiana 
University, 141-143. 

Bernardo, John J. and Lawrence X. Tarpey, Sr. (1980), "Brand Loyalty as 
a Theoretical Construct," in Lamb and Dunne Theoretical 
Developments in Marketing , eds. Charles W. Lamb, Jr. and Patrick 
M. Dunne, Chicago, IL: American Marketing Association, 198-201. 

Best, Arthur (1981), When Consumers Complain , Xew York, NY: Columbia 
University Press. 

Biehal, Gabriel J. (1983), "Consumer's Prior Experiences and 

Perceptions in Auto Repair Choice," Journal of Marketing 47 
(Summer), 82-91 . 

Bjorklund, Gail and Richard Bjorklund (1979), "An Exploratory Study of 
Toddler's Satisfaction With Their Toy Environments," in Advances 
in Consumer Research: Volume 6 , ed. William L. Wilkie, Ann Arbor. 
MI: Association for Consumer Research, 400-406. 

Bloch, Peter H. (1981), "An Exploration into the Scaling of Consumers' 
Involvement With a Product Class," in Advances in Consumer 
Research: Volume 8 , ed. Kent B. Monroe, Ann Arbor, MI: 
Association for Consumer Research, 61-65. 



417 



, and G. D. Bruce (1984), "The Leisure Experience and 

Consumer Products: An Investigation of Underlying Satisfactions," 
Journal of Leisure Research 16, 74-88. 

, and Marsha L. Richins (1983), "A Theoretical Model for 



the Study of Product Importance Perceptions," Journal of Marketing 
47 (Summer), 69-81 . 

[loom, Paul N. (1982a), "Research on Consumerism: Opportunities and 
Challenges," in Advances in Consumer Research: Volume 9 , ed. 
Andrew Mitchell, Ann Arbor, MI: Association for Consumer Research, 
520-522. 

(1982b), Consumerism and Beyond: Research Perspectives on 



the Future Social Environment , Cambridge, MA: Marketing Science 
Institute . 

, and Stephen A. Greyser (1981), Exploring the Future of 



Consumerism , Cambridge, MA: Marketing Science Institute. 

Bohl, Don Lee (ed.) (1987), Close to the Customer: An American 

Management Association Research Report on Consumer Affairs , New 
York, NY: American Management Association. 

Bolfing, Claire P., and Robert B. Woodruff (1987), "Effects of 
Situational Involvement on Consumers ' Use of Standards in 
Satisfaction/Dissatisfaction Processes," paper presented at 10th 
Consumer Satisfaction/Dissatisfaction and Complaining Behavior 
Conference, Montreal, Quebec, August. 

Braden, Marcia and Elaine Walster (1964), "The Effect of Anticipated 

Dissonance on Pre-Decision Behavior," in Leon Festinger Conflict, 
Decision and Dissonance , Stanford, CA: Stanford University Press, 
145-151 . 

Brehm, Jack W. (1956), "Postdecision Changes in the Desirability of 

Alternatives," Journal of Abnormal and Social Psychology 52 (May) 
384-389. 

■(1959), "Increasing Cognitive Dissonance by a Fait 



Accompli," Journal of Abnormal and Social Psychology 58 (May), 
379-382. 

and Arthur R. Cohen (1962), Explorations in Cognitive 



Dissonance , New York, NY: John Wiley & Sons Inc. 

Srewer, Marilynn B. (1977), "An Information Processing Approach to 
Attribution of Responsibility," Journal of Experimental Social 
Psychology 13, 53-69. 



418 



Brown, Stephen W. and Teresa A. Swartz (1984), "Consumer Medical 
Complaint Behavior: Determinants of and Alternatives to 
Malpractice Litigation," Journal of Public Policy and Marketing 3, 
85-98. 

Buchanan, Bruce, Moshe Givon and Arieh Goldman (1987), "Measurement of 
Discrimination Ability in Taste Tests: An Empirical Investiga- 
tion," Journal of Marketing Research 24 (May), 154-163. 

3urke, Marian, W. David Conn, and Richard J. Lutz ( 1 978a) ,_ "Using 
Psychographic Variables to Investigate Product Disposition 
Behaviors," Working Paper #64, Center for Marketing Studies, UCLA. 

, W. David Conn and Richard J. Lutz (1978b), "Using 



Psychographic Variables to Investigate Product Disposition 
Behaviors," in Research Frontiers in Marketing: Dialogues and 
Directions , 1978 Educator's Proceedings Series #43, ed. Subhash C. 
Jain, Chicago, IL: AiMA, 321-326. 

Burnkrant, Robert E. and Alain Cousineau (1975), "Informational and 

Normative Social Influence in Buyer Behavior," Journal of Consumer 
Research 2 (December), 206-215. 

, and Alan G. Sawyer (1983), "Effects of Involvement and 



Message Content on Information Processing Intensity, in 
Information Processing Research in Advertising , ed. Richard J. 
Harris, Hillsdale, NJ: Lawrence Erlbaum Assoc, 43-64. 

Cadotte, Ernest R. , Robert B. Woodruff, and Roger L. Jenkins ( 1 987) , 
"Expectations and Norms in Models of Consumer Satisfaction," 
Journal of Marketing Research 24 (August), 305-314. 

, Robert B. Woodruff, and Roger L. Jenkins (1983), "Norms 

and Expectation Predictions: How Different Are the Measures?" in 

International Fare in Consumer Satisfaction and Complaining 
Behavio r' eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: 
Indiana University, 49-56. 

Calder, Bobby J. (1973), "Cognitive Consistency and Consumer Behavior," 
in Perspectives in Consumer Behavior , revised edition, eds. Harold 
H. Kassarjian and Thomas S. Robertson, Glenview, IL: Scott, 
Foresman and Co., 247-263. 

Cardozo, Richard N. (1965), "An Experimental Study of Customer Effort, 
Expectation and Satisfaction," Journal of Marketing Research 2 
(August), 244-249. 

Carey, J. Ronald, Steven H. Clicque, Barbara A. Leighton, and Frank^ 
Milton (1976), "A Test of Positive Reinforcement of Customers," 
Journal of Marketing 40 (October), 98-100. 



419 



Carlsmith, J. Merrill and Elliot Aronson (1963), "Some Hedonic 
Consequences of the Confirmation and Disconf irmation of 
Expectancies," Journal of Abnormal and Social Psychology 66 
(Number 2), 151-156. 

and Jonathan L. Freedman (1968), "Bad Decisions and 



Dissonance: Nobody's Perfect," in Theories of Cognitive 
Consistency: A Sourcebook , eds . Robert P. Abelson et al., 
Chicago, IL: Rand McNally and Company, 485-490. 

Carmines, Edward G. and Richard Z. Zeller (1979), Reliability and 

Validity Assessment , Sage University Paper series on Quantitative 
Analysis in the Social Sciences, 07-017, Beverly Hills, CA: Sage 
Publications . 

Carroll, Stephen J. and Henry L. Tosi (1977), Organizational Behavior , 
Chicago, IL: St. Clair Press. 

Carsky, Mary L. (1987), "Negative Information at the Point-of- Sale: 
Can It Moderate Satisfaction with New Automobiles?" Paper 
presented at 10th Consumer Satisfaction/ Dissatisfaction and 
Complaining Behavior Conference, Montreal, Quebec, August. 

Chakravarti, Dipankar and John G. Lynch (1983), "A Framework for 

Exploring Context Effects on Consumer Judgment and Choice," in 
Advances in Consumer Research: Volume 10 , eds. Richard P. Bagozzi 
and Alice M. Tybout, Ann Arbor, MI: Association for Consumer 
Research, 289-297. 

Charell, Ralph (1973), How I Turn Ordinary Complaints into Thousands of 
Dollars , New York, NY: Dell Publishing Co. 

Chase, William G. and K. Anders Ericsson (1982), "Skill and Working 

Memory," in The Psychology of Learning and Motivation: Volume 16 , 
ed. Gordon H. Bower, New York, NY: Academic Press, 1-58. 

and Herbert A. Simon (1973), "Perception in Chess," 



Cognitive Psychology , 4 (January), 55-81. 

Churchill, Gilbert A. Jr. (1979), "A Paradigm for Developing Better 

Measures of Marketing Constructs," Journal of Marketing Research 
16 (February), 64-73. 

Churchill, Gilbert A. Jr. and Anthony Pecotich (1982), "A Structural 
Equation Investigation of the Pay Satisfaction - Valence 
Relationship Among Salespeople," Journal of Marketing 46 (Fall), 
114-124. 

, and Carol Surprenant (1932), "An Investigation into the 



Determinants of Customer Satisfaction," Journal of Marketing 
Research 19 (November), 491-504. 



420 



Cialdini, Robert B., John T. Cacioppo, Rodney Bassett, and John A. 
Miller (1978), "Low-Ball Procedure for Producing Compliance: 
Commitment Then Cost," Journal of Personality and Social 
Psychology 36 (Number 5), 463-476. 

Clarke, Keith and Russell W. Belk (1979), "The Effects of Product 

Involvement and Task Definition on Anticipated Consumer Effort," 
in Advances in Consumer Research: Volume 6 , ed. William L. Wilkie, 
Ann Arbor, MI: Association for Consumer Research, 313-318. 

Cobb, Cathy J., Gary C. Walgren, and Mary Hollowed (1987), "Differences 
in Organizational Responses to Consumer Letters of Satisfaction 
and Dissatisfaction," in Advances in Consumer Research: Volume 14 , 
eds. Melanie Wallendorf and Paul Anderson, Provo, UT: Association 
for Consumer Research, 227-231 . 

Coca-Cola (1982), "Measuring the Grapevine: Consumer Response and 

Word-of-Mouth," study conducted by Technical Assistance Research 
Programs, Inc. 

Cohen, Joel B. (1981), "The Role of Affect in Categorization: Toward a 
Reconsideration of the Concept of Attitude," in Advances in 
Consumer Research: Volume 9 , ed. Andrew Mitchell, Ann Arbor, MI: 
Association for Consumer Research, 94-100. 

(1982), "Involvement: Separating the State from Its Causes 

and Effects," Working Paper #33, Center for Consumer Research, 
University of Florida. 

(1983), "Involvement and You: 1000 Great Ideas," Advances 
in Consumer Research: Volume 10 , eds. Richard P. Bagozzi and Alice 
M. Tybout, Ann Arbor, MI: Association for Consumer Research, 
324-327. 

and Marvin E. Goldberg (1970), "The Dissonance Model in 

Post-Decision Product Evaluation," Journal of Marketing Research 7 
(August), 315-321 . 

and Michael J. Houston (1972), "Cognitive Consequences of 

Brand Loyalty," Journal of Marketing Research 9 (February), 97-99. 

Converse, Philip E. (1970), "Attitudes and Non-attitudes: Continuation 
of a Dialogue," in The Quant i tative Analysis of Social Problems , 
ed. Edward R. Tufte, Reading, MA: Addison-Wesley , 168-189. 

(1974), "Comment: The Status of Nonatt itudes , " .American 



Political Science Review 68, 650-666. 

3ook, Stuart W. and Claire Selltiz (1964), "A Multiple Indicator 
Approach to Attitude Measurement," reprinted in Readings in 
Attitude Theory and Measurement , ed. Martin Fishbein, New York, 
NY: John Wiley & Sons, 1967, 220-235. 



421 



Cook, Thomas D. and Donald T. Campbell (1979), Quasi -Experiment at ion: 
Design and Analysis Issues for Field Settings , Boston, MA: 
Houghton Miffflin Company. 

Costley, Carolyn L. (1988), "Meta Analysis of Involvement Research," in 
Advances in Consumer Research , Volume 15, ed. Michael J. Houston, 
Provo, UT: Association for Consumer Research, 554-562. 

Crosby, Lawrence A. and James R. Taylor (1983), "Psychological 
Commitment and Its Effects on Post-Decision Evaluation and 
Preference Stability Among Voters," Journal of Consumer Research 9 
(March), 413-431 . 

, and James R. Taylor (1982), "Consumer Satisfaction with 

Michigan's Container Deposit Law - An Ecological Perspective," 
Journal of Marketing 46 (Winter), 47-60. 

Cummings, William H. and M. Venkatesan (1976), "Cognitive Dissonance 
and Consumer Behavior: A Review of the Evidence," Journal of 
Marketing Research 13 (August), 303-308. 

Cureton, Edward E. (1950), "Validity, Reliability and Baloney," 
reprinted in Principles of Educational and Psychological 
Measurement , eds. William A. Mehrens and Robert L. Ebel, Chicago, 
IL: Rand McNally, 1967, 271-273. 

Czepiel, John A. (1980). Managing Customer Satisfaction in Consumer 
Service Businesses , Cambridge, MA: Marketing Science Institute. 

and Larry J. Rosenberg (1977), "The Study of Consumer 



Satisfaction: Addressing the 'So What' Question, in 
Conceptualization and Measurement of Consumer Satisfaction and 
Dissatisfaction , ed. H. Keith Hunt, Cambridge, MA: Marketing 
Science Institute, 92-119. 

Davidson, Jon R. and Sara B. Kiesler (1964), "Cognitive Behavior Before 
and After Decisions," in Leon Festinger, Conflict, Decision and 
Dissonance , Stanford, CA: Stanford University Press, 10-21. 

Day, Ralph L. (1977), "Extending the Concept of Consumer 

Satisfaction," in Advances in Consumer Research: Volume 4 , ed. 
William D. Perreault, Jr., Atlanta: Association for Consumer 
Research, 1 49-1 54 . 

(1980a), "How Satisfactory Is Research on Consumer 



Satisfaction?" in Advances in Consumer Research: Volume 7 , ed. 
Jerrv C. Olson, Ann Arbor, MI: Association for Consumer Researcl 
593-597. 



422 



(1980b), "Research Perspectives on Consumer Complaining 



Behavior," in Theoretical Developments in Marketing , eds. Charles 
W. Lamb, Jr. and Patrick M. Dunne, Chicago, IL: American Marketing 
Association, 211-215. 

(1983), "The Next Step: Commonly Accepted Constructs for 



Satisfaction Research," in International Fare in Consumer 
Satisfaction and Complaining Behavior , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 113-117. 

(1984), "Modeling Choices Among Alternative Responses to 



Dissatisfaction," Advances in Consumer Research: Volume 11 , ed , 
Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 
496-499. 

and Stephen B. Ash (1979), "Consumer Response to 



Dissatisfaction With Durable Products," in Advances in Consumer 
Research: Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: 
Association for Consumer Research, 438-444. 

and Muzaffer Bodur (1978), "Consumer Response to 



Dissatisfaction With Services and Intangibles," in Advances in 
Consumer Research: Volume 5 , ed. H. Keith Hunt, Ann Arbor, MI: 
Association for Consumer Research, 263-272. 

and Muzaffer Bodur (1979), "Analysis of Average 



Satisfaction Scores of Individuals Over Product Categories, in 
New Dimensions of Consumer Satisfaction and Complaining Behavior , 
eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana 
University, 184-189. 

and E. Laird Landon, Jr. (1976), "Collecting Comprehensive 



Consumer Complaint Data by Survey Research," in Advances in 
Consumer Research: Volume 3 , ed. Beverlee B. Anderson, Ann Arbor, 
MI: Association for Consumer Research, 263-268. 

DeBell, Margaret and Rachel Dardis (1979), "Extending Product Life: 
Technology Isn't the Only Issue," in Advances in Consumer 
Research: Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: 
Association for Consumer Research, 381-385. 

Deighton, John (1984), "The Interaction of Advertising and Evidence," 
Journal of Consumer Research 11 (December), 763-770. 

Dickson, Peter R. and William L. Wilkie (1978), The Consumption of 

Household Durables: A Behavioral Review , Cambridge, MA: Marketing 
Science Institute. 



423 



Diener, Betty J. (1980), "Selective Perception and Consumer 

Dissatisfaction (Or... The Case of the Rabid Rabbits)," in Refining 
Concepts and Measures of Consumer Satisfaction and Complaining 
Behavior , eds . H. Keith Hunt and Ralph L. Day, Bloomington, IN: 
Indiana University, 168-172. 

(1977), "The Fulfillment of Consumer Promotions as a 

Source~of Consumer Dissatisfaction," in Conceptualization and 
Measurement of Consumer Satisfaction and Dissatisfaction , ed. H. 
Keith Hunt, Cambridge, MA: Marketing Science Institute, 333-354. 

Donnelly, James H. Jr., and John M. Ivancevich (1970), "Post-Purchase 
Reinforcement and Back-out Behavior," Journal of Marketing 
Research 7 (August), 399-400. 

Eastlack, J.O. Jr. (1964), "Consumer Flavor Preference Factors in Food 
Product Design," Journal of Marketing Research 1 (February), 
38-42. 

Edwards, Allen L. and Kathryn Claire Kenney (1946), "A Comparison of 
the Thurstone and Likert Techniques of Attitude Scale 
Construction," reprinted in Readings in Attitude Theory and 
Measurement , ed. Martin L. Fishbein (1967), New York, NY: John 
Wiley & Sons, 249-256. 

Eisenberger, Kenneth (1977), The Expert Consumer: A Complete Handbook , 
Englewood Cliffs, NJ : Prentice-Hall. 

Engel, James F. , Robert J. Kegerreis, and Roger D. Blackwell (1969), 
"Word-of-Mouth Communication by the Innovator," Journal of 
Marketing 33 (July), 15-19. 

Evans, Richard H. (1983), "Diagnosing Customer Satisfaction/ 

Dissatisfaction and Behavioral Intention: Expectancy Value Theory 
Versus Equity Theory," in International Fare in Consumer 
Satisfaction and Complaining Be havior , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 64-68. 

Fazio, Russell H. , Jeaw-mei Chen, Elizabeth C. McDonel, and Steven J. 
Sherman (1982), "Attitude Accessibility, Attitude-Behavior 
Consistency, and the Strength of the Object-Evaluation 
Association," Journal of Experimental Social Psychology 13, 
339-357. 

, Martha C. Powell, and Paul M. Herr (1983), "Toward a 

Process Model of the Attitude-Behavior Relation: Accessing One's 
Attitude upon Mere Observation of the Attitude Object," Journal of 
Personality and Social Psychology 44 (Number 4), 723-735. 



424 



, David M. Sanbonmatsu, Martha C. Powell, and Frank R. 

Kardes (1986), "On the Automatic Activation of Attitudes," Journal 
of Personality and Social Psychology 50 (Number 2), 229-238. 

, and Mark P. Zanna (1981), "Direct Experience and 

Attitude-Behavior Consistency," in Advances in Experimental Social 
Psychology: Volume 14 , ed. Leonard Berkowitz, New York, NY: 
Academic Press, 161-202. 

Feick, Lawrence F. (1984), "Do Consumers Complain Sequentially?" in 
Proceedings of 1984 Winter Educator's Conference: Scientific 
Method in Marketing , eds. Paul F. Anderson and Michael J. Ryan, 
Chicago, IL: American Marketing Association, 89-92. 

(1987), "Latent Class Models for the Analysis of 

Behavioral Hierarchies," Journal of Marketing Research 24 (May), 
174-186. 

Feinman, Jeffrey (1979), The Purple Pages , New York, NY: Hawthorn 
Books, Inc. 

Feldman, Daniel C. and Hugh J. Arnold (1983), Managing Individual and 
Group Behavior in Organizations , New York, NY: McGraw-Hill Book 
Co. 

Fendrich, James M. (1967), "A Study of the Association Among Verbal 

Attitudes, Commitment and Overt Behavior in Different Experimental 
Situations," reprinted in The Consistency Controversy , ed. Allen 
E. Liska, New York, NY: John Wiley & Sons, 1972, 95-108. 

Festinger, Leon (1957), A Theory of Cognitive Dissonance , Evanstcn, IL: 
Row, Peterson and Co. 

(1964), Conflict, Decision and Dissonance, Stanford, CA: 



Stanford University Press. 

Filer, Robert J. (1952), "Frustration, Satisfaction, and Other Factors 
Affecting the Attractiveness of Goal Objects," Journal of Abnormal 
and Social Psychology 47, 203-212. 

Fishbein, Martin (1972), "The Search for Attitudinal-Behavioral 

Consistency," reprinted in Perspectives in Consumer Behavior , 
Revised Edition eds. Harold H. Kassarjian and Thomas S. Robertson, 
Glenview, IL: Scott, Foresman and Co, 1973, 210-219. 

(1966), "The Relationship Between Beliefs, Attitudes and 

Behaviors," reprinted in Research in Consumer Behavior , eds. David 
T. Kollat, Roger D. Blackwell and James F. Engel, New York, NY: 
Holt, Rinehart and Winston, 1970, 216-235. 



425 



and leek Ajzen (1975), Belief, Attitude, Intention and 

Behavior: An Introduction to Theory and Research , Reading, MA: 
Add i son-Wesley. 

Fisk, Raymond P. and Clifford E. Young (1985), "Disconf irmation of 
Equity Expectations: Effects on Consumer Satisfaction With 
Services," in Advances in Consumer Research, Volume 12 , eds.^ 
Elizabeth Hirschman and Morris B. Holbrook, Provo, UT: Association 
for Consumer Research, 340-345. 

Folkes, Valerie S. (1984a), "An Attributional Approach to Postpurchase 
Conflict Between Buyers and Sellers," Advances in Consumer 
Research: Volume 11 , ed. Thomas C. Kinnear, Provo, UT: Association 
for Consumer Research, 500-503. 

(1984b), "Consumer Reactions to Product Failure: An 

Attributional Approach," Journal of Consumer Research 10 (March), 
398-409. 

, Susan Koletsky, and John L. Graham (1987), "A Field Study 

of Causal Inferences and Consumer Reaction: The View from the 
Airport," Journal of Consumer Research 13 (March), 534-539. 

, and Barbara Kotsos (1986), "Buyers' and Sellers' 

Explanations for Product Failure: Who Done It?" Journal of 
Marketing 50 (April), 74-80. 

Fornell, Claes and Nicholas Didow (1980), "Economic Constraints on 

Consumer Complaining Behavior," in Advances in Consumer Research: 
Volume 7 , ed. Jerry C. Olson, Ann Arbor, MI: Association for 
Consumer Research, 318-323. 

and Birger Wernerfelt (1987), "Defensive Marketing 

Strategy by Customer Complaint Management: A Theoretical 
Analysis,"' Journal of Marketing Research 24 (November) 337-346. 

and Robert A. Westbrook (1979), "An Exploratory Study ^ of 

Assertiveness, Aggressiveness, and Consumer Complaining Behavior," 
Advances in Consumer Research: Volume 6 , ed. William L. Wilkie, 
Ann Arbor, MI: Association for Consumer Research, 105-110. 

and Robert A. Westbrook (1984), "The Vicious Circle^of 

Consumer Complaints," Journal of Marketing 48 (Summer), 68-78. 

Freedman, Jonathan L. (1964), "Involvement, Discrepancy and Change," 

Journal of Abnormal and Social Psychology 69 (September), 290-295. 

Futrell, Charles M. and A. Parasuraman (1984), "The Relationship of 

Satisfaction and Performance to Salesforce Turnover," Journal of 
Marketing 48 (Fall), 33-40. 



426 



Gerard, Harold B. (1965), "Deviation, Conformity, and Commitment," in 
Current Studies in Social Psychology , eds. Ivan D. Steiner and 
Martin Fishbein, New York, NY: Holt Rinehart and Winston, Inc., 
263-277. 

(1968), "Basic Features of Commitments," in Theories of 



C ognitive Consistency: A Sourcebook , (eds. Robert P. Abelson et 
al., Chicago, IL: Rand McNally and Company, 456-463. 

Gibson, Lawrence D. (1981), "The Psychology of Food: Why We Eat What We 
Eat When We Eat It," Food Technology (February), 54-56. 

Gilly, Mary C, William L. Cron, and Thomas E. Barry (1983), "The 

Expectations-Performance Comparison Process: An Investigation of 
Expectation Types," in International Fare in Consumer Satisfaction 
and Complaining Behavior , eds. Ralph L. Day and H. Keith Hunt, 
Bloomington, IN: Indiana University, 10-16. 

and Betsy D. Gelb (1982), "Post-Purchase Consumer 



Processes and the Complaining Consumer," Journal of Consumer 
Research 9 (December), 323-328. 

Goering, Patricia A. (1985), "Effects of Product Trial on Consumer 

Expectations, Demand and Prices," Journal of Consumer Research 12 
(June) , 7 4-82. 

Goodwin, Stephen A., Vijay Mahajan, and Bhal J. Bhatt (1979), "On 

Consumer Dissatisfaction: Consumer Arbitration as an Alternative 
Dispute Resolution Mechanism," Advances in Consumer Research: 
Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: Association for 
Consumer Research, 1979, 460-465. 

Gorn, Gerald J. (1982), "The Effects of Music in Advertising on Choice 
Behavior: A Classical Conditioning Approach," Journal of 
Marketing 46 (Winter), 94-101. 

and Clark Leavitt (1984), "Audience Involvement in 
Advertising: Four Levels," Journal of Consumer Research 11 (June), 
581-592. 

Greenwald, Anthony G. and Clark Leavitt (1984), "Audience Involvement 
in Advertising: Four Levels," Journal of Consumer Research 11 
(June), 581-592. 

Gronhaug, Kjell (1977), "Exploring Consumer Complaining Behavior: A 

Model and Some Empirical Results," Advances in Consumer Research: 
Volume 4 , ed. William D. Perreault, Jr., Atlanta: Association for 
Consumer Research, 159-165. 



427 



and Johan Arndt (1980), "Consumer Dissatisfaction and 

Complaining Behavior as Feedback: A Comparative Analysis of Public 
and Private Delivery Systems," Advances in Consumer Research: 
Volume 7 , ed. Jerry C. Olson, Ann Arbor, MI: Association for 
Consumer Research, 324-328. 

and Gerald Zaltman (1981), "Complainers and Noncomplainers 



Revisited: Another Look at the Data," Advances in Consumer 
Research: Volume 8 , ed. Kent B. Monroe, Ann Arbor, MI: 
Association for Consumer Research, 83-87. 

Grubb, Edward L. and Harrison L. Grathwohl (1967), "Consumer 

Self-Concept, Symbolism and Market Behavior: A Theoretical 
Approach," Journal of Marketing 31 (October), 22-27. 

Gurwitz, Sharon B. and Kenneth A. Dodge (1977), "Effects of 
Confirmations and Disconf irmations on Stereotype-Based 
Attributions," Journal of Personality and Social Psychology 35, 
495-500. 

Guttman, Louis and Edward A. Suchman (1947), "Intensity and a Zero 

Point for Attitude Analysis," reprinted in Readings in Attitude 
Theory and Measurement , ed. Martin Fishbein, New York, NY: John 
Wiley & Sons, 1967, 267-276. 

Haines, George H. Jr. (1979), "Three Papers on Consumer 

Satisfaction/Dissatisfaction: A Comment," in Advances in Consumer 
Research: Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: 
Association for Consumer Research, 450-452. 

Hartley, Eugene L. (1967), "Attitude Research and the Jangle Fallacy," 
in Attitude, Ego-Involvement and Change , eds . Carolyn W. Sherif 
and Muzafer Sherif, New York, NY: John Wiley & Sons, 88-104. 

Havlena, William J. and Morris B. Holbrook (1986), "The Varieties of 
Consumption Experience: Comparing Two Typologies of Emotion in 
Consumer Behavior," Journal of Consumer Research 13 (December), 
394-404. 

Hawes, Douglass K. , Roger D. Blackwell, and W. Wayne Talarzyk (1976), 

"Consumer Decisions to Reduce or Stop Using Products and Services: 
Preliminary Results of a Nationwide Study," Advances in Consumer 
Research: Volume 3 , ed. Beverlee B. Anderson, Ann Arbor, MI; 
Association for Consumer Research, 102-109. 

Heberlein, Thomas A., Daniel Linz, and Bonnie P. Ortiz (1982), 
"Satisfaction, Commitment and Knowledge of Customers on a 
Mandatory Participation Time-of-Day Electricity Pricing 
Experiment," Journal of Consumer Research 9 (June), 106-114. 



428 



Heller, Harry E. (1982), "If You Want to Know What is Important, Don't 
Ask," in Straight Talk About Attitude Research , ed. Joseph Chasm, 
Chicago, IL: American Marketing Association, 1-11. 

Helson, Harry (1948), "Adaptation-Level as a Basis for a Quantitative 
Theory of Frames of Reference," Psychological Review 55 
(November), 297-313. 

Herndon, Booton (1972), Satisfaction Guaranteed: An Unconventional 
Report to Today's Consumers , New York, New York: McGraw-Hill. 

Herzberg, Frederick, Bernard Mausner, and Barbara B. Snyderman (1959), 
The Motivation to Work , Second edition, New York, NY: John Wiley & 
Sons, Inc. 

Hoch, Stephen J. and Young-Won Ha (1986), "Consumer Learning: 

Advertising and the Ambiguity of Product Experience," Journal of 
Consumer Research 13 (September), 221-233. 

Horowitz, David (1979), Fight Back, and Don't Get Ripped Off , San 
Francisco, CA: Harper & Row. 

Houston, Michael J. and Michael L. Rothschild (1978a), "AParadigm for 
Research on Consumer Involvement," Working Paper, University of 
Wisconsin-Madison. 

(1978b), "Conceptual and Methodological Perspectives on 

Involvement," in Research Frontiers in Marketing: Dialogues and 
Direction s, ed. Subhash C. Jain, Chicago, IL: American Marketing 
Association, 184-187. 

Hover, Wayne D. (1984), "An Examination of Consumer Decision Making for 
a Common Repeat Purchase Product," Journal of Consumer Research 11 
(December), 822-829. 

Hughes, Donald A. (1977), "An Investigation of the Relation of Selected 
Factors to Consumer Satisfaction," in Conceptualization and 
Measure ment of Consumer Satisfaction and Dissatisfaction , ed. H. 
Keith Hunt, Cambridge, MA: Marketing Science Institute, 300-332. 

Hunt, H. Keith (1976), "CS/D: The Program Planning and Evaluation 

Perspective," in Advances in Consumer Research: Volume 3 , ed. 
Beverlee 3. Anderson, Ann Arbor, MI: Association for Consumer 
Research, 259-260. 

ed. (1977a), Conceptualization and Measurement of 
Consumer Satisfaction and Dissatisfaction , "Introduction" and 
"CS/D — Overview and Future Research Directions," Cambridge, MA: 
Marketing Science Institute. 



429 



(1977b), "CS/D: Bits and Pieces," in Consumer 

Satisfaction, Dissatisfaction and Complaining Behavior , ed. Ralph 
L. Day, Bloomington, IN: Indiana University Conference April 
1977, '38-41. 

, (1982), "Consumer Satisfaction/Dissatisfaction: Of 

Continuing Interest? To Whom? Why?" in Consumerism and Beyond: 
Research Persectives on the Future Social Environment , ed. Paul N. 
Bloom, Cambridge, MA: Marketing Science Institute. 

(1983), "A CS/D & CB Bibliography— 1 982, " in 



"" International Fare in Consumer Satisfaction and Complaining 

Behavior , eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: 
Indiana University, 132-154. 

Hunt, Shelby D. (1976), Marketing Theory: Conceptual Foundations of 
Research in Marketing , Columbus, OH: Grid, Inc. 

Huppertz, John W. (1979), "Measuring Components of Equity in the 

Marketplace: Perceptions of Inputs and Outcomes by Satisfied and 
Dissatisfied Consumers," in New Dimensions of Consumer 
Satisfaction and Complaining Behavior , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 140-143. 

, Sidney J. Arenson, and Richard H. Evans (1978), "An 

Application of Equity Theory to Buyer-Seller Exchange Situations," 
Journal of Marketing Research 15 (May), 250-260. 

Isen, Alice M. , Margaret Clark, and Mark F. Schwartz (1976), "Duration 
of the Effect of Good Mood on Helping: 'Footprints on the Sands of 
Time'," Journal of Personality and Social Psychology 34 (Number 
3), 385-393. 

Izard, Carroll E. (1972), Patterns of Emotions: A New Analysis of 
Anxiety and Depression , New York, NY: Academic Press. 

(1977), Human Emotions, New York, NY: Plenum Press. 



Jacoby, Jacob and Robert W. Chestnut (1978), Brand Loyalty: Measurement 
and Management , New York, NY: John Wiley & Sons. 

and James J. Jaccard (1981), "The Sources, Meaning and 

Validity of Consumer Complaint Behavior: A Psychological 
Analysis," Journal of Retailing 57 (Fall), 4-24. 

James, William (1890), The Principles of Psychology , Volumes 1 and 2, 
reprint editions, New York, NY: Dover Publications, Inc. 

Jensen, Michael C. and William H. Meckling (1976), "Theory of the Firm: 
Managerial Behavior, Agency Costs and Ownership Structure," 
Journal of Financial Economics 3 (October), 305-360. 



430 



Johnson, Michael P. (1973), "Commitment: A Conceptual Structure and 
Empirical Application," The Sociological Quarterly 14 (Summer), 

395-406. 

Jordan, Kathi A. and James H. Leigh (1984), "Race and Sex Differences 
of Children in Satisfaction and Other Factors Associated With 
Video Games," in Advances in Consumer Research, Volume 11 , ed. 
Thomas Kinnear, Ann Arbor, MI: Association for Consumer Research, 
94-99. 

Kahneman, Daniel and Amos Tversky (1973), "On the Psychology of 
Prediction," Psychological Review 80 (Number 4), 237-251. 

Kamins, Michael A. and Henry Assael (1987), "Two-Sided Versus One-Sided 
Appeals: A Cognitive Perspective on Argumentation, Source 
Derogation and the Effect of Disconf irming Trial on Belief 
Change," Journal of Marketing Research 24 (February), 29-39. 

Kassarjian, Harold H. and Thomas S. Robertson, eds. (1973), 

Perspectives in Consumer Behavior , Revised Edition, Glenview, IL: 
Scott, Foresman and Co. 

Kennedy, John R. and Peter C. Thirkell (1987), "An Extended Perspective 
on the Antecedents of Satisfaction," paper presented at 10th 
Consumer Satisfaction/Dissatisfaction and Complaining Behavior 
Conference, Montreal, Quebec, August. 

and Peter C. Thirkell (1982), "The Effect of Automobile 

Product Experience on Attribute Disconf irmation and Importance," 
in New Findings on Consumer Satisfaction and Complaining , eds. 
Ralph L. Day and H. Keith Hunt, Bloomington, IN: Indiana 

University, 2-10. 

Keppel, Geoffrey (1982), Design and Analysis: A Researcher's Handbook , 
Englewood Cliffs, NJ: Prentice-Hall. 

Kiesler, Charles A. (1971), The Psychology of Commitment: Experiments 
Linking Behavior to Belief , New York, NY: Academic Press. 

and Michael S. Pallak (1976), "Arousal Properties of 

Dissonance Manipulations," Psychological Bulletin 83 (Number 6), 
1014-1025. 

Klemke, Lloyd W. (1982), "Exploring Juvenile Shoplifting," Sociology 
and Social Research , 67, 59-75. 

Knauer, Virginia H. (1983), Consumer's Resource Handbook , Pueblo, CO: 
Consumer Information Center. 

Koepp, Stephen (1985), "All Wrapped Up in Company Logos," Time 
(December 2) , 70 . 



431 



Kourilsky, Marilyn and Trudy Murray (1981), "The Use of Economic 

Reasoning to Increase Satisfaction with Family Decision Making," 
Journal of Consumer Research 8 (September), 183-188. 

Krishnan, S. and Valerie A. Valle (1979), "Dissatisfaction Attributions 
and Consumer Complaint Behavior," Advances in Consumer Research: 
Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: Association for 
Consumer Research, 445-449. 

Kunin, Theodore (1955), "The Construction of a New Type of Attitude 
Measure," Personnel Psychology 8 (Spring), 65-77. 

LaBarbera, Priscilla A. and David Mazursky (1983), "A Longitudinal 

Assessment of Consumer Satisfaction/ Dissatisfaction: The Dynamic 
Aspect of the Cognitive Process," Journal of Marketing Research 20 
(November), 393-404. 

Lachman, Roy, Janet L. Lachman and Earl C. Butterfield (1979), 

Cognitive Psychology and Information Processing: An Introduction , 
Hillsdale, NJ : Lawrence Erlbaum Associates. 

Laird, James D. (1974), "Self-Attribution of Emotion: The Effects of 

Expressive Behavior on the Quality of Emotional Experience," 

Journal of Personality and Social Psychology 29 (Number 4), 
475-486. 

Lambert, David R. , Ronald J. Dornoff, and Jerome B. Kernan (1977), "The 
Industrial Buyer and the Postchoice Evaluation Process," Journal 
of Marketing Research 14 (May), 246-251. 

Lmgmeyer, Lynn and George Miaoulis (1981), "The Consumer and the 

Health Care Proces," in Advances in Consumer Research: Volume 7, 



ed. Jerrv C. Olson, Ann Arbor, MI: Association for Consumer 
Research', 612-615. 

LaPiere, Richard T. (1934), "Attitudes Vs. Actions," Social Forces 13 
(Number 2), 230-237. 

Lastovicka, John L. and E. H. Bonfield (1982), "Do Consumers Have Brand 
Attitudes?" Journal of Economic Psychology 2, 57-75. 

LaTour, Stephen A. (1978), "Determinants of Participant and Observer 
Satisfaction With Adversary and Inquisitorial Modes of 
Adjudication," Journal of Personality and Social Psychology 36 
(Number 1 2 ) , 1 53 1 - 1 545 . 

and Nancy C. Peat (1978), "Effects of Prior Experience and 

Quality of the Best Alternative on Consumer Satisfaction, 
Complaints and Purchase Termination Responses to Quality Decline," 
Working Paper #5, University of Florida Center for Consumer 
Research . 



432 



and Nancy C. Peat (1979), "Conceptual and Methodological 

Issues in Consumer Satisfaction Research," Advances in Consumer 
Research: Volume 6 , ed. William L. Wilkie, Ann Arbor, MI: 
Association for Consumer Research, 431-437. 

and Nancy C. Peat (1980), "The Role of Situationally 



Produced Expectations, Other's Experiences and Prior Experience in 
Determining Consumer Satisfaction," Advances in Consumer Research: 
Volume 7 , ed. Jerry C. Olson, Ann Arbor, MI: Association for 
Consumer Research, 588-592. 

Leavitt, Clark (1976), "Consumer Satisfaction as the Ultimate Life 

Force," in Advances in Consumer Research: Volume 3 , ed. Beverlee 
B. Anderson, Ann Arbor, MI: Association for Consumer Research, 
252-258. 

(1977), "Consumer Satisfaction and Dissatisfaction: 



Bi-Polar or Independent?" in Conceptualization and Measurement of 
Consumer Satisfaction and Dissatisfaction , ed. H. Keith Hunt, 
Cambridge, MA: Marketing Science Institute, 132-149. 

Leigh, James H. (1984), "What Buyers Prefer the Next Time Around," 
Tierra Grande (Number 24), 12-13. 

Levy, Danna and Carol Surprenant (1982), "A Comparison of Responses co 
Dissatisfaction with Products and Services," in Conceptual and 
Empirical Contributions to Con sumer Satisfaction and Complainin g 
Behavior , eds. H. Keith Hunt and Ralph L Day, Bloomington, IN: 
Indiana University, 43-49. 

Lewin, Kurt (1951), Field Theory in Social Science , New York, NY: 
Harper. 

(1952), "Group Decision and Social Change," in Readings in 

Social Psychology , eds. G. Swanson, T. Newcomb and E. Hartley, New 
York, NY: Holt. 

Lewis, Robert C. (1983), "Consumers Complain - What Happens When 

Business Responds?" in International Fare in Consumer Satisfaction 
and Complaining Behavior , eds. Ralph L. Day and H. Keith Hunt, 
Bloomington, IN: Indiana University, 88-94. 

Liefeld, John (1980), "Urban/Rural Consumer Expectations and 

Evaluations of Their Consumer Realities," in Refining Concepts and 
Measures of Consumer Satisfaction and Complaining Behavior , eds. 
H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana 
University, 95-101. 



433 



Likert, Rensis (1932), "The Method of Constructing an Attitude Scale," 
reprinted in Readings in Attitude Theory and Measurement , ed. 
Martin Fishbein, New York, NY: John Wiley & Sons, 1967, 90-95. 

Lippert, Alice and Maurice MacDonald (1981), "Evaluating User 

Satisfaction With a Consumer Protection Hotline," Journal of 
Consumer Affairs 15 (Summer), 136-145. 

Locke, Edwin A. (1967), "Relationship of Success and Expectation to 

Affect on Goal-Seeking Tasks," Journal of Personality and Social 
Psychology 7 (Number 2), 125-134. 

(1969), "What Is Job Satisfaction?," Organizational 



Behavior and Human Performance 4 (November), 309-336. 

Locker, David and David Dunt (1978), "Theoretical and Methodological 
Issues in Sociological Studies of Consumer Satisfaction With 
Medical Care," Social Science and Medicine 12 (July), 283-292. 

Lundstrom, William J. and Laurence M. Lamont (1976), "The Development 
of a Scale to Measure Consumer Discontent," Journal of Marketing 
Research 13 (November), 373-381. 

Lutz, Richard J. (1980), "On Getting Situated: The Role of 

Situational Factors in Consumer Research," Advances in Consumer 
Research: Volume 3 , ed. Beverlee B. Anderson, Ann Arbor, MI: 
Association for Consumer Research, 659-663. 

(1975), "Changing Brand Attitudes Through Modification of 

Cognitive Structure," Journal of Consumer Research 1 (March), 49- 
59. 

Lynch, John G. Jr. (1984a), "Comments on 'Intentions and Behaviors'," 
Advances in Consumer Research: Volume 11 , ed. Thomas C. Kinnear, 
Provo, UT: Association for Consumer Research, 156-158. 

(1984b), "Adventures in Paramorphic Modeling: Models of 



Consumers' Processing of Negative Information," Working Paper 
University of Florida Center for Consumer Research. 

Lyon, Don and Paul Slovic (1976), "Dominance of Accuracy Information 
and Neglect of Base Rates in Probability Estimation," Acta 
Psychologica 40, 287-298. 

MacCorquodale, Kenneth and Paul E. Meehl (1948), "On a Distinction 
Between Hypothetical Constructs and Intervening Variables," 
Psychological Review 55 (Number 1), 95-107. 

MacDougall, Curtis D. (1958), Hoaxes , New York, NY: Dover Publications. 



434 



Maddox, R. Neil (1981), "Two-Factor Theory and Consumer Satisfaction: 

Replication and Extension," Journal of Consumer Research 8 (June), 
97-102. 

Marks, Lawrence J., Susan Higgins, and Michael A. Kamins (1988), 

"Investigating the Experiential Dimensions of Product Evaluation," 
Advances in Consumer Research: Volume 15 , ed. Michael J. Houston, 
Provo, UT: Association for Consumer Research, 114-121. 

Maslow, A. H. (1943), "A Theory of Human Motivation," Psychological 
Review 50, 370-396. 

Maxwell, John C, Jr. (1988), "Coffee Consumption Continues to Cool 
Off," Advertising Age , (June 13), 14. 

McNeal, James U. (1969), "Consumer Satisfaction: The Measure of 

Marketing Effectiveness," MSU Business Topics (Summer), 31-35. 

McQuarrie, Edward F. and J. Michael Munson (1987), "The Zaichkowsky 
Personal Involvement Inventory. Modification and Extension," in 
Advances in Consumer Research: Volume 14 , Melanie Wallendorf and 
Paul Anderson (eds.), Provo, UT, Association for Consumer 
Research, 36-40. 

and J. Michael Munson (undated), "A Multi- dimensional, 

Arousal-Based Measure of Consumer Involvement," Unpublished 
working paper, Department of Marketing, Santa Clara University, 
Santa Clara, CA 95053. 

Mehrabian, Albert and James A. Russell (1974), An Approach to 
Environmental Psychology , Cambridge, MA: MIT Press. 

Miller, John A. (1977), "Studying Satisfaction, Modifying Models, 

Eliciting Expectations, Posing Problems, and Making Meaningful 
Measurements," in Conceptualization and Measurement of Consumer 
Satisfaction and Dissatisfaction , ed. H. Keith Hunt, Cambridge, 
MA: Marketing Science Institute, 72-91. 

Mills, Michael K. (1981), "Deviance and Dissatisfaction: An Exploratory 
Study," Advances in Consumer Research: Volume 8 , ed. Kent B. 
Monroe, Ann Arbor, MI: Association for Consumer Research, 
682-686. 

Mitchell, Andrew A. (1979), "Involvement: A Potentially Important 

Mediator of Consumer Behavior," in Advances in Consumer Research: 
Volume 6 , ed . William L. Wilkie, Ann Arbor, MI: Association for 
Consumer Research, 191-196. 



435 



Mittal, Bonwari and Myung-Sou Lee (1988), "Separating Brand-Choice 
Involvement from Product Involvement Via Consumer Involvement 
Profiles," in Advances in Consumer Research , Volume 15, ed. 
Michael J. Houston, Provo, UT : Association for Consumer Research, 
43-49. 

Mizerski, Richard W. , Jon B. Freiden and Robert C. Greene, Jr. (1983), 
"The Effect of the 'Don't Know' Option on TV Ad Claim Recognition 
Tests," in Advances in Consumer Research: Volume 10 , eds. Richard 
P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: Association for 
Consumer Research, 283-287. 

Mook, Douglas G. (1987), Motivation: The Organization of Action , New 
York, NY: W.W. Norton and Company. 

Moore, Danny L. , Douglas Hausknecht, and Kanchana Thamodaran (1986), 

"Time Compression Response Opportunity and Persuasion," Journal of 
Consumer Research 13 (June), 85-99. 

Moore, Ellen M. and F. Kelly Shuptrine (1984), "Disconf irmation Effects 
on Consumer Satisfaction and Decision Making Processes," Advances 
in Consumer Research: Volume 11 , ed. Thomas C. Kinnear, Provo, 
UT: Association for Consumer Research, 299-304. 

Morgan, Fred W. and Dana I. Avrunin (1982), "Consumer Conduct in 

Product Liability Litigation," Journal of Consumer Research 9 
(June), 47-55. 

Morganoskv, Michelle Ann and Hilda Mayer Buckley (1987), "Complaint 
Behavior: Analvsis by Demographics, Lifestyle and Consumer 
Values," Advances in Consumer Research: Volume 14 , eds. Melanie 
Wallendorf and Paul Anderson, Provo, UT: Association for Consumer 
Research, 223-226. 

Morris, Earl W. (1977), "A Normative Deficit Approach to Consumer 

Satisfaction," in Conceptualization and Measurement of Consumer 
Satisfaction and Dissatisfaction , ed. H. Keith Hunt, Cambridge, 
MA: Marketing Science Institute, 240-274. 

, Mary Winter, and Ivan F. Beutler (1976), "A Normative 



Deficit Model of Consumer Behavior," in Advances in Consumer 
Research : Volume 3 , ed. Beverlee B. Anderson, Ann Arbor, MI: 
Association for Consumer Research, 161-165. 

Moschis, George P. (1976), "Social Comparison and Informal Group 

Influence," Journal of Marketing Research 13 (August), 237-24^ 

, Dena Saliagas Cox, and James J. Kellaris (1987), "An 



Exploratory Study of Adolescent Shoplifting Behavior," in Advances 

in Consumer Research: Volume 14 , eds. Melanie Wallendorf and Paul 

Anderson, Provo, UT: Association for Consumer Research, 526-530. 



436 



Mowen, John C. and Stephen J. Grove (1983), "Search Behavior, Price 
Paid, and the 'Comparison Other': An Equity Theory Analysis of 
Post Purchase Satisfaction," in International Fare in Consumer 
Satisfaction and Complaining Behavior , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 57-63. 

Muncy, James A. and Shelby D. Hunt (1984), "Consumer Involvement: 
Definitional Issues and Research Directions," in Advances in 
Consumer Research: Volume 11 , ed. Thomas C. Kinnear, Provo, UT: 
Association for Consumer Research, 193-196. 

Narayana, Chem L. and Rom J. Markin (1975), "Consumer Behavior and 

Product Performance: An Alternative Conceptualization," Journal of 
Marketing 39 (October), 1-6. 

Neisser, Ulric (1967), Cognitive Psychology , New York, NY: Meredith 
Publishing. 

Norpoth, Helmut and Milton Lodge (1985), "The Difference between 

Attitudes and Nonattitudes in the Mass Public: Just Measurements?" 
American Journal of Political Science 29, 291-307. 

Nunnally, Jum C. (1978), Psychometric Theory , second edition, New York, 
NY: McGraw-Hill. 

Ofir, Chezy (1982), "Judgment Under Uncertainty: The Integration of 

Singular and Distributional Information," Working Paper Number 32, 
University of Florida Center for Consumer Research. 

Olander, Folke (1980), "Recent Developments in European Consumer Policy 
Research," in Advances in Consumer Research: Volume 7 , ed. Jerry 
C. Olson, Ann Arbor, MI: Association for Consumer Research, 
56-65. 

(1977), "Consumer Satisfaction — A Skeptic's View," in 

Conceptualization and Measurement of Consumer Satisfaction and 
Dissatisfaction , ed. H. Keith Hunt, Cambridge, MA: Marketing 
Science Institute, 409-452. 

Oliva, Terence A. and Alvin C. Burns (1978), "Catastrophe Theory as a 
Model for Describing Consumer Behavior," Advances in Consume r 
Research: Volume 5 , ed. H. Keith Hunt, Ann Arbor, MI: Association 
for Consumer Research, 273-276. 

Oliver, Richard L. (1977), "Effect of Expectation and Disccnf irmat ion 
on Postexposure Product Evaluations: An Alternative 
Interpretation," Journal of Applied Psychology 62 (Number 4), 
480-486. 

(1979), "An Interpretation of the Attitudinal and 
Behavioral Effects of Puffery," Journal of Consumer Affairs 13 
(Summer) , 8-27 . 



4 37 



(1980a), "A Cognitive Model of the Antecedents and 
Consequences of Satisfaction Decisions," Journal of Marketing 
Research 17 (November), 460-469. 

(1980b), "Conceptualization and Measurement of 



in 



Disconf irmation Perceptions in the Prediction of Consumer 
Satisfaction," in Refining Concepts and Measures of Consumer 
Satisfaction and Complaining Behavior , eds. H. Keith Hunt and 
Ralph L. Day, Bloomington, IN: Indiana University, 2-6. 

(1980c), "Theoretical Bases of Consumer Satisfaction 

Research: Review, Critique and Future Direction," in Charles W. 
Lamb, Jr. and Patrick M. Dunne (eds.) Theoretical Developments 
Marketing , Chicago, IL: American Marketing Association, 206-210. 

(1981), "Measurement and Evaluation of Satisfaction 
Processes in Retail Settings," Journal of Retailing 57 (Fall), 
25-48. 

(1987), "An Investigation of the Interrelationship 
Between Consumer (Dis) Satisfaction and Complaint Reports," 
Advances in Consumer Research: Volume 14 , eds. Melanie Wallendorf 
and Paul Anderson, Provo, UT: Association for Consumer Research, 
218-222. 

and William 0. Bearden (1983), "The Role of Involvement in 

Satisfaction Processes," Advances in Consumer Research: Volume 10 . 
eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: 
Association for Consumer Research, 250-255. 

and William 0. Bearden (1985), "Disconf irmation Processes 

and Consumer Evaluations in Product Usage," Journal of Business 
Research 13 (June), 235-246. 

and Wayne S. DeSarbo (1988), "Response Determinants in 

Satisfaction Judgments," Journal of Consumer Research 14 (March), 
495-507. 

and Gerald Linda (1981), "Effect of Satisfaction and^Its 

Antecedents on Consumer Preference and Intention," Advances in 
Consumer Research: Volume 8 , ed. Kent B. Monroe, Ann Arbor, MI: 
Association for Consumer Research, 88-93. 

and Robert A. Westbrook (1982), "The Factor Structure of 

Satisfaction and Related Postpurchase Behavior," in New Findings 
on Consumer Satisfaction and Complaining , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 11-14. 

Olshavsky, R. W. and John A. Miller (1972), "Consumer Expectations, 

Product Performance and Perceived Quality," Journal of Marketing 
Research 9 ( February), 19-21. 



438 



Olson, Jerry C. and Philip A. Dover (1976), "Effects of Expectation 
Creation and Disconf irmation on Belief Elements of Cognitive 
Structure," Advances in Consumer Research: Volume 3 , ed. Beverlee 
B. Anderson, Ann Arbor, MI: Association for Consumer Research, 
168-175. 

and Philip A. Dover (1978), "Cognitive Effects of 



Deceptive Advertising," Journal of Marketing Research 15 
(February) , 29-38. """" 

Orne, Martin T. (1962), "On the Social Psychology of the Psychological 
Experiment: With Particular Reference to Demand Characteristics 
and Their Implications," American Psychologist 17, 776-783. 

Ortinau, David J. (1982), "Testing the Influence of Time/Usage 
Experience on Consumers' Perceived, Expected and Actual 
Satisfaction Attitudes Toward Automobiles," in New Findings on 
Consumer Satisfaction and Complaining , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 23-28. 

Oshikawa, Sadaomi (1969), "Can Cognitive Dissonance Theory Explain 
Consumer Behavior?" Journal of Marketing 33 (October), 44-49. 

Park, C. Whan and William J. Bahr (1980), "A Situational Analysis of 

Communication Effect: A New Product Purchase by Mail Order," in 

Advances in Consumer Research: Volume 7 , ed. Jerry C. Olson, Ann 
Arbor, MI: Association for Consumer Research, 650-654. 

and V. Parker Lessig (1977), "Students and Housewives: 



Differences in Susceptibility to Reference Group Influence," 
Journal of Consumer Research 4 (September), 102-110. 

Pe-due, Barbara C. and John 0. Summers (1986), "Checking the Success of 
Manipulations in Marketing Experiments," Journal of Marketing 
Research 23 (November), 317-325. 

Peter, J. Paul (1984), "What a MESS: Comments on Materialism, 

Ethnocentrism and Symbolism Session," in Advances in Consumer 
Research: Volume 1 1 , ed. Thomas C. Kinnear, Provo, UT: 
Association for Consumer Research, 298. 

and Lawrence Tarpey (1975), "A Comparative Analysis of 



Three Consumer Decision Strategies," Journal of Consumer Resear ch 
2 (June), 29-37. 

Peterson, Robert A., Wayne D. Hoyer and William R. Wilson (1986), The 

Role of Affect in Consumer Behavior , Lexington, MA: D.C. Heath and 
Company. 

Petty, Richard E. and John T. Cacioppo (1986), Communication and 

Persuasion: Central and Peripheral Routes to Attitude Change , New 
York, NY: Springer-Verlag . 



439 



, John T. Cacioppo and David Schumann (1983), "Central and 

Peripheral Routes to Advertising Effectiveness: The Moderating 
Role of Involvement," Journal of Consumer Research 10 (September), 
135-146. 

Pfaff, Martin (1977), "The Index of Consumer Satisfaction: Measurement 
Problems and Opportunities," in Conceptualization and Measurement 
of Consumer Satisfaction and Dissatisfaction , ed. H. Keith Hunt, 
Cambridge, MA: Marketing Science Institute, 36-72. 

Pierce, John C. and Douglas D. Rose (1974), "Nonattitudes and American 
Public Opinion: The Examination of a Thesis," American Political 
Science Review 68, 626-649. 

Prakash, Ved and John W. Lounsbury (1983), "A Reliability Problem in 
the Measurement of Disconf irmation of Expectations," in Advances 
in Consumer Research: Volume 10 , eds. Richard P. Bagozzi and Alice 
M. Tybout, Ann Arbor, MI: Association for Consumer Research, 
244-249. 

Pucely, Marya J., Richard Mizerski, and Pamela Perrewe (1988), "A 
Comparison of Involvement Measures for the Purchase and 
Consumption of Pre-Recorded Music," in Advances in Consumer 
Research , Volume 15, ed. Michael J. Houston, Provo, UT: 
Association for Consumer Research, 37-42. 

Reitz, H. Joseph (1981), Behavior in Organizations , revised edition, 
Homewood, IL: Richard D. Irwin, Inc. 

Resnik, Alan, Brian Gnauck and Rodney Aldrich (1977), "Corporate 

Responsiveness to Consumer Complaints," in Consumer Satisfaction, 
Dissatisfaction and Complaining Behavior , ed. Ralph L. Day, 
Bloomington, IN: Indiana University, 148-152. 

and Robert R. Harmon (1983), "Consumer Complaints and 

Managerial Response: A Holistic Approach," Journal of Marketing 47 
(Winter), 86-97. 

Reymert, Martin L. , ed. (1928), Feelings and Emotions: The Wittenberg 
Symposium , New York, NY: Arno Press. 

Rhine, Ramon J. and Laurence J. Severance (1970), "Ego-Involvement, 

Discrepancy, Source Credibility, and Attitude Change," Journal of 
Personality and Social Psychology 16 (No. 2), 175-190. 

Richins, Marsha L. (1987), "A Multivariate Analysis of Responses to 

Dissatisfaction," Journal of the Academy of Marketing Science 15 
(Fall), 24-31. 



440 



(1985), "Factors Affecting the Level of Consumer Initiated 

Complaints to Marketing Organizations," in Consumer Satisfaction, 
Dissatisfaction and Complaining Behavior , eds. H. Keith Hunt and 
Ralph L. Day, Bloomington, IN: Indiana University, 82-85. 

(1983a), "Negative Word-of-Mouth by Dissatisfied 

Consumers: A Pilot Study," Journal of Marketing 47 (Winter), 
63-78. 

(1983b), "Word-of-Mouth as an Expression of Product 

Dissatisfaction," in International Fare in Consumer Satisfaction 
and Complaining Behavior , eds. Ralph L. Day and H. Keith Hunt, 
Bloomington, IN: Indiana University, 100-104. 

(1982), "An Investigation of Consumers' Attitudes Toward 

Complaining," Advances in Consumer Research: Volume 9 , ed. Andrew 
Mitchell, Ann Arbor, MI: Association for Consumer Research, 
502-506. 

and Peter H. Bloch (1987), "The Role of Situational and 

Enduring Involvement in Post-Purchase Product Evaluation," Paper 
presented at 10th Consumer Satisfaction/Dissatisfaction and 
Complaining Behavior Conference, Montreal, Quebec, August. 

and Peter H. Bloch (1986), "After the New Wears Off: The 



Temporal Context of Product Involvement," Journal of Consumer 
Research 13 (September), 280-285. 

, and Teri Root-Shaffer (1988), "The Role of Involvement 



and Opinion Leadership in Consumer Word-of-Mouth: An Implicit 
Model Made Explicit," in Advances in Consumer Research , Volume 15. 
ed. Michael J. Houston, Provo, UT: Association for Consumer 
Research, 32-36. 

Risley, George (1972), Modern Industrial Marketing , New York, NY: 
McGraw-Hill . 

Robertson, Thomas S. (1976), "Low-Commitment Consumer Behavior," 
Journal of Advertising Research 16 (April), 19-24. 

, John R. Rossiter, and Scott Ward (1985), "Consumer 



Satisfaction Among Children," in Advances in Consumer Research, 
Volume 12 , eds. Elizabeth C. Hirschman and Morris B. Holbrook, 
Provo, UT: Association for Consumer Research, 279-284. 

Robinson, John A. (1980), "Affect and Retrieval of Personal Memories,' 
Motivation and Emotion 4 (Number 2), 149-174. 

(1976), "Sampling Autobiographical Memory," Cognitive 



Psychology 8, 578-595. 



441 



Robinson, Larry M. and Robert L. Berl (1980), "What About Compliments: 
A Follow-Up Study on Customer Complaints and Compliments," in 
Refining Concepts and Measures of Consumer Satisfaction and 
Complaining Behavior , eds. H. Keith Hunt and Ralph L. Day, 
Bloomington, IN: Indiana University, 144-148. 

, George G. Trebbi, and Roy D. Adler (1983), "Taking It to 



the Top: A Profile of Customers Who Complain to the CEO, in 
International Fare in Consumer Satisfaction and Complaining 
Behavior , eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: 
Indiana University, 95-99. 

Rodgers, Alice L. and Daniel J. Sweeney (1980), "Satisfaction With 

Retail Stores as Reflected in Consumers' Opinions," in Refining 
Concepts and Measures of Consumer Satisfaction and Complaining 
Behavior , eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: 
Indiana University, 153-158. 

Rokeach, Milton (1970), Beliefs, Attitudes and Values , San Francisco, 
CA: Jossey-Bass, Inc. 

Rosander, A. C. (1937), "An Attitude Scale Based Upon Behavior 
Situations," Journal of Social Psychology 8, 3-15. 

Rosenthal, Robert (1982), "Valid Interpretation of Quantitative 

Research Results," in New Directions for Methodology of Social and 
Behavioral Science: Forms of Validity in Research , eds. David 
Brinberg and L. Kidder, San Francisco, CA: Jossey-Bass, 59-75. 

Ross, Ivan (1985), "Consumer Initiated Communication Audit," in 

Consumer Satisfaction, Dissatisfaction, and Complaining Behavior , 
eds. H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana 
University, 73-81 . 

and Kathy Gardner (1985), "The Use of Consumer Initiated 



Communication as Marketing Research Data: The View of Consumer 
Affairs Professionals," unpublished manuscript, University of 
Minnesota. 

Ross, Stephen A. (1973), "The Economic Theory of Agency: The 

Principal's Problem," American Economic Review 63 (May), 134-139. 

Rothschild, Michael L. (1984), "Perspectives on Involvement: Current 
Problems and Future Directions," Advances in Consumer Research: 



Volume 11 , ed. Thomas C. Kinnear, Provo, UT: Association for 
Consumer Research, 216-217. 

Ruekert, Robert W. and Gilbert A. Churchill, Jr. (1984), "Reliability 
and Validity of Alternative Measures of Channel Member 
Satisfaction," Journal of Marketing Research 21 (May), 226-233. 



442 



Rusbult, Caryl E. (1983), "A Longitudinal Test of the Investment Model: 
The Development (and Deterioration) of Satisfaction and Commitment 
in Heterosexual Involvements," Journal of Personality and Social 
Psychology 45 (Number 1), 101-117. 

and Isabella Zembrodt (1983), "Responses to 

Dissatisfaction in Romantic Involvements: A Multidimensional 
Scaling Analysis," Journal of Experimental Social Psychology 19, 
274-293. 

Russo, J. Edward (1979), "Consumer Satisfaction/Dissatisfaction: An 

Outsider's View," in Advances in Consumer Research: Volume 6 , ed. 
William L. Wilkie, Ann Arbor, MI: Association for Consumer 
Research, 453-455. 

Sampson, Edward A. and Linda B. Sibley (1965), "A Further Examination 
of the Confirmation or Nonconf irmation of Expectancies and 
Desires," Journal of Personality and Social Psychology 2 (Number 
1), 133-137. 

Sawyer, Alan G. (1975), "Demand Artifacts in Laboratory Experiments in 
Consumer Research," Journal of Consumer Research 1 (March), 20-30. 

(1977), "The Role of Role Playing in Experiments About 

Consumer Behavior," in Contemporary Marketing Thought , eds. 
Barnett A. Greenberg and Danny N. Bellenger, Chicago, IL: American 
Marketing Association, 191-194. 

, Parker M. Worthing, and Paul E. Sendak (1979), "The Role 

of Laboratory Experiments to Test Marketing Strategies," Journal 
of Marketing 43 (Summer), 60-67. 

Scammon, Debra L. and Mary Jane Sheffet (1984), "Regulation in the _ 
80' s: What is the Role of Consumer Researchers?" in Advances in 
Consumer Research: Volume 11 , ed. Thomas C. Kinnear, Provo, UT: 
Association for Consumer Research, 463-465. 

Schachter, Stanley (1964), "The Interaction of Cognitive and 

Physiological Determinants of Emotional State," in Advances in 
Experimental Social Psychology , ed. Leonard Berkowitz, New York, 
NY: Academic Press, 49-80. 

Scherf, Gerhard W.H. (1974), "Consumer Dissatisfaction as a Function ci 
Dissatisfaction with Interpersonal Relationships," Journal of 
Applied Psychology 59 (Number 4), 465-471. 

Schmalensee, Richard (1978), "A Model of Advertising and Product 

Quality," Journal of Political Economy 86 (Number 3), 485-503. 

Scott, Carol A. (1976), "The Effects of Trial and Incentives on Repeat 
.Purchase Behavior," Journal of Marketing Research 13 (August), 
263-269. 



443 



and Richard F. Yalch (1980), "Consumer Response to Initial 

Product Trial: A Bayesian Analysis," Journal of Consumer Research 
7 (June), 32-41 . 

Sechrest, Lee (1979), Unobtrusive Measurement Today , San Francisco, CA: 
Jossey-Bass, Inc. 

Serafin, Raymond (1987), "Auto Makers Stress Consumer Satisfaction," 
Advertising Age , February 23, S-12. 

Settle, Robert B. (1973), "Attribution Theory and Consumer Behavior," 
in Perspectives in Consumer Behavior , revised edition eds. Harold 
H. Kassarjian and Thomas S. Robertson, Glenview, IL: Scott, 
Foresman and Co., 64-74. 

Shapiro, Benson P. (1968), "The Psychology of Pricing," Harvard 
Business Review 46 (July-August), 14, 25, 160. 

Shaw, Marvin E. and Philip R. Costanzo (1982), Theories of Social 
Psychology , second edition, New York, NY: McGraw-Hill. 

and Jack M. Wright (1967), Scales for the Measurement of 



Attitudes , New York, NY: McGraw-Hill. 

Shelly, Maynard W. , ed. (1972) Analyses of Satisfaction , Volumes I, II, 
III, New York, NY: MSS Information Corporation. 

Sherif, Muzafer and Carl I. Hovland (1961), Social Judgment: 

Assimilation and Contrast Effects in Communication and Attitude 
Change , New Haven, CT: Yale University Press. 

Sherman, Elaine and Leon G. Schiffman (1984), "Applying Age-Gender 
Theory from Social Gerontology to Understand the Consumer 
Well-Being of the Elderly," in Advances in Consumer Research: 
Volume 11 , ed. Thomas C. Kinnear, Provo, UT: Association for 
Consumer Research, 569-573. 

Sherrell, Daniel L. and Terence A. Shimp (1982), "Consumer Involvement 
in a Laboratory Setting," in An Assessment of Marketing Thought 
and Practice , ed. Bruce J. Walker, William 0. Bearden, William R. 
Darden, Patrick E. Murphy, John R. Nevin, Jerry C. Olson, Barton 
A. Weitz, Chicago, IL: American Marketing Association, 104-108. 

Sheth, Jagdish N. (1971), "Word-of-Mouth in Low-Risk Innovations," 
Journal of Advertising Research 11 (June), 15-18. 

Shuptrine, F. Kelly and Gerhard Wenglorz (1981), "Comprehensive 

Identification of Consumers' Marketplace Problems and What They Do 
About Them," in Advances in Consumer Research, Volume VIII , ed. 
Kent Monroe, Ann Arbor, MI: Association for Consumer Research, 
687-692. 



444 



and Gerhard Wenglorz (1980), "Product Satisfaction/ 

Dissatisfaction: How Content Are Consumers with Products They Buy 
and Use?" in Refining Concepts and Measures of Consumer 
Satisfaction and Comp laining Beha vior , eds . H. Keith Hunt and 
Ralph L. Day, Bloomington, IN: Indiana University, 149-152. 

Shuv-Ami, Avichai (1984), "Measures of Disconf irmation: An 

Integration," in Proceedings of 1984 AMA Winter Educators' 
Conference: Scientific Method in Marketing , eds. Paul F. Anderson 
and Michael J. Ryan, Chicago, IL: American Marketing Association, 
233-237. 

Silverman, Irwin (1968), "Expectancy Disconf irmation and the Choice of 
Negative Alternatives: Dissonance Avoidance or Situational 
Demands?" in Theories of Cognitive Consistency: A Sourcebook , 
(eds. Robert P. Abelson et al., Chicago, IL: Rand McNally and 
Company, 479-484. 

Singh, Jagdip (1988), "Consumer Complaint Intentions and Behavior: 
Definitional and Taxonomical Issues," Journal of Marketing 52 
(January), 93-107. 

Sirgy, M. Joseph (1982), "Self-Concept in Consumer Behavior: A Critical 
Review," Journal of Consumer Research 9 (December), 287-300. 



(1984), "A Social Cognition Model of Consumer 

Satisfaction/Dissatisfaction," Psychology and Marketing 1 (Number 
2, Summer) , 27-44 . 

Smead, Raymond J., James B. Wilcox, and Robert E. Wilkes (1980), "An 
Illustration and Evaluation of a Joint Process Tracing 
Methodology: Eye Movement and Protocols," Advances in Consumer 
Research: Volume 7 , ed. Jerry C, Olson, Ann Arbor, MI: 
Association for Consumer Research, 507-512. 

Smith, Darlene B. and Paul N. Bloom (1984), "Is Consumerism Dead or 

Alive? Some New Evidence," Advances in Consumer Research: Volu me 
JJ_> ed. Thomas C. Kinnear, Provo, UT: Association for Consumer 
Research, 369-373. 

Smith, Patricia C, Lome M. Kendall, and Charles L. Hulin (1969), The 
Measurement of Satisfaction in Work and Retirement: A Strategy for 
the Study of Attitudes , Chicago, IL: Rand McNally and Co. 

Smith, Robert E. and William R. Swinyard (1982), "Information Response 
Models: An Integrated Approach," Journal of Marketing 46 
(Winter), 81-93. 



445 



and William R. Swinyard (1983), "Attitude-Behavior 

Consistency: The Impact of Product Trial Versus Advertising," 
Journal of Marketing Research 20 (August), 257-262. 

Sobel, Robert S. and Howard McGuire (1977), "Subtractive Models of 
Satisfaction: Cognitive Processes or Measurement Artifact?" 
Journal of Psychology 96 (July), 251-258. 

Solomon, Michael R. (1986), "The Missing Link: Surrogate Consumers in 
the Marketing Chain," Journal of Marketing 50 (October), 208-218. 

Solomon, Richard L. and John D. Corbit (1974), "An Opponent-Process 
Theory of Motivation: I. Temporal Dynamics of Affects," 
Psychological Review 81 (March), 119-145. 

Steers, Richard M. and Lyman W. Porter (1975), Motivation and Work 
Behavior , New York, NY: McGraw-Hill. 

Stein, Karen F. (1980), "Explaining Ghetto Consumer Behavior: 

Hypotheses from Urban Sociology," Journal of Consumer Affairs 14 
(Summer), 232-241 . 

Sternthal, Brian, Alice M. Tybout and Bobby J. Calder (1987), 

"Confirmatory Versus Comparative Approaches to Judging Theory 
Tests," Journal of Consumer Research 14 (June), 114-125. 

Stone, Robert N. (1984), "The Marketing Characteristics of 

Involvement," in Advances in Consumer Research: Volume 11 , ed. 
Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 
210-215. 

Strumpel, Burkhard, ed. (1974), Subjective Elements of Weil-Being 
Paris: Organisation for Economic Co-operation and Development. 

Summers, John 0. and Donald H. Granbois (1977), "Predictive and 

Normative Expectations in Consumer Dissatisfaction and Complaining 
Behavior," in Advances in Consumer Research: Volume 4 , ed. William 
D. Perreault, Jr. , Atlanta: Association for Consumer Research, 
155-158. 

Surprenant, Carol and Gilbert A. Churchill, Jr. (1984), "Can Role^ 
Playing be Substituted for Actual Consumption?" in Advances in 
Consumer Research: Volume 11 , ed. Thomas C. Kinnear, Provo, UT: 
Association for Consumer Research, 122-126. 

Swan, John E. (1983), "Consumer Satisfaction Research and Theory: _ 
Current Status and Future Direction," in International Fare in 
Consumer Satisfaction and Complaining Behavior , eds. Ralph L. Day 
and H. Keith Hunt, Bloomington, IN: Indiana University, 124-129. 



446 



(1987), "Consumer Satisfaction Related to 



Disconfirmation of Expectations and Product Performance," Paper 
presented at 10th Consumer Satisfaction/ Dissatisfaction and 
Complaining Behavior Conference, Montreal, Quebec, August 1987. 

and Maxwell G. Carroll (1980), "Patient Satisfaction: An 



Overview of Research 1965-1978," in Refining Concepts and Measures 
of Consumer Satisfaction and Complaining Behavior , eds. H. Keith 
Hunt and Ralph L. Day, Bloomington, IN: Indiana University, 
112-118. 

and Linda Jones Combs (1976), "Product Performance and 



Consumer Satisfaction: A New Concept," Journal of Marketing 40 
(April), 25-33. 

and Warren S. Martin (1981), "Testing Comparison Level and 



Predictive Expectations Models of Satisfaction," Advances in 
Consumer Research: Volume 8 , ed. Kent B. Monroe, Ann Arbor, MI: 
Association for Consumer Research, 77-82. 

and I. Frederick Trawick (1979), "Testing an Extended 



Concept of Consumer Satisfaction," in New Dimensions of Consumer 
Satisfaction and Complaining Behavior , eds. Ralph L. Day and H. 
Keith Hunt, Bloomington, IN: Indiana University, 56-61 . 

and I. Frederick Trawick (1983), "Satisfaction With and 



Support of a Continuously Utilized Service: A Public Radio 
Station," in International Fare in Consumer Satisfaction and 
Complaining Behavior , eds. Ralph L. Day and H. Keith Hunt, 
Bloomington, IN: Indiana University, 75-80. 

and I. Frederick Trawick (1980a), "Satisfaction Related 



Predictive Vs. Desired Expectations," in Refining Concepts and 
Measures of Consumer Satisfaction and Complaining Behavior , eds. 
H. Keith Hunt and Ralph L. Day, Bloomington, IN: Indiana 
University, 7-12. 

and I. Frederick Trawick (1980b), "Triggering Cues and the 



Evaluation of Products as Satisfactory or Dissatisfactory," in 
Proceedings of the 1979 Educators' Conference , Chicago, IL: 
American Marketing Association, 231-234. 

, I. Frederick Trawick, and Maxwell G. Carroll (1981), 



"Effect of Participation in Marketing Research on Consumer 
Attitudes Toward Research and Satisfaction with a Service," 
Journal of Marketing Research 18 (August), 356-363. 

Tapp, Shelley R. (1984), "Brunswik's Lens Model: A Review and Extension 
to Consumer Research," in Advances in Consumer Research: Volume 
1 1 , ed. Thomas C. Kinnear, Provo, UT: Association for Consumer 
Research, 103-108. 



447 



Technical Assistance Research Programs, Inc. (1979), Consumer Complaint 
Handling in America: Final Report , Washington, DC: TARP. 

Temerlin, Maurice Kahn (1963), "On Choice and Responsibility in a 
Humanistic Psychotherapy," Journal of Humanistic Psychology 3, 
35-48. 

Triandis, Harry C. and Leigh M. Triandis (1965), "Some Studies of 

Social Distance," reprinted in Readings in Attitude Theory and 
Measurement , ed. Martin Fishbein, New York, NY: John Wiley & Sons, 
1967, 199-207. 

Trombetta, William L. and Timothy L. Wilson (1975), "Foreseeability of 
Misuse and Abnormal Use of Products by the Consumer," Journal of 
Marketing 39 (July), 48-55. 

Tse, David K. and Peter C. Wilton (1988), "Models of Consumer 

Satisfaction Formation: An Extension," Journal of Marketing 
Research 25 (May), 204-212. 

Valle, Valerie and Melanie Wallendorf (1977), "Consumers' Attributions 
of the Cause of Their Product Satisfaction and Dissatisfaction, 
in Consumer Satisfaction, Dissatisfaction and Complaining 
Behavior , ed. Ralph L. Day, Bloomington, IN: Indiana University, 
26-30. 

Vredenburg, Harrie and Chow-Hou Wee (1986), "The Role of Customer 
Service in Determining Customer Satisfaction," Journal of the 
Academy of Marketing Science 14 (Summer), 17-26. 

Vroom, Victor H. (1964), Work and Motivation New York, NY: John Wiley & 
Sons . 

Wallendorf, Melanie and Eric J. Arnould (1988), "My Favorite Things: A 
Cross-Cultural Inquiry into Object Attachment, Possessiveness , and 
Social Linkage," Journal of Consumer Research 14 (March), 531-547. 

, Russell Belk, and Deborah Heisley (1988), "Deep Meaning 

in Possessions: The Paper," in Advances in Consumer Research , 
Volume 15, ed. Michael J. Houston, Provo, UT: Association for 
Consumer Research, 528-530. 

Walster, Elaine (1964), "The Temporal Sequence of Post-Decision 
Processes," reprinted in Readings in Social Psychology , eds. 
Jonathan L. Freedman, J. Merrill Carlsmith and David 0. Sears, 
Englewood Cliffs, NJ: Prentice-Hall, 1971, 517-528. 

Walton, John R. and Eric N. Berkowitz (1985), "Consumer Decision Making 
and Perceived Decision Freedom," in Advances in Consumer Research , 
Volume 12, eds. Elizabeth C. Hirschman and Morris B. Holbrook, 
Provo, UT: Association for Consumer Research, 461-464.5. 



448 



Wanous, John P. and Edward E. Lawler III (1972), "Measurement and 
Meaning of Job Satisfaction," Journal of Applied Psychology 56 
(Number 2), 95-105. 

Watts, William A. (1968), "Predictability and Pleasure: Reactions to 
the Disconfirmation of Expectancies," in Theories of Cognitive 
Consistency: A Sourcebook , (eds.) Robert P. Abelson et al., 
Chicago, IL: Rand McNally and Company, 456-463. 

Webb, Eugene J., Donald T. Campbell, Richard D. Schwartz, and Lee 

Sechrest (1966), Unobtrusive Measures: Nonreactive Research in the 
Social Sciences , Chicago, IL: Rand McNally College Publishing Co. 

Weick, Karl G. (1967), "Promise and Limitations of Laboratory 

Experiments in the Development of Attitude Change Theory," in 
Carolyn W. Sherif and Muzafer Sherif, eds., Attitude, 
Ego-Involvement and Change , New York, NY: John Wiley & Sons, 
51-75. 

Weiner, Bernard (1985), Human Motivation , New York, NY: Springer- 
Verlag. 

(1980), "A Cognitive (Attribution)-Emotion-Action Model of 

Motivated Behavior: An Analysis of Judgments of Help-Giving," 
Journal of Personality and Social Psychology 39 (Number 2), 
186-200. 

, Dan Russell, and David Lerman (1979), "The 



Cognition-Emotion Process in Achievement-Related Contexts, 
Journal of Personality and Social Psychology , 37 (Number 7), 
1211-1220. 

, Dan Russell and David Lerman (1978), "Affective 



Consequences of Causal Ascriptions," in New Directions in 
Attribution Research, Volume 2 , eds. John H. Harvey, William Ickes 
and Robert F. Kidd, Hillsdale, NJ: Lawrence Erlbaum Associates, 
59-90. 

Westbrook, Robert A. (1977), "A Study of Consumer Dissatisfaction 

Before Purchase," in Advances in Consumei Research: Volume 4 , ed. 
William D. Perreault, Jr., Atlanta: Association for Consumer 
Research, 142-148. 

(1980a), "Intrapersonal Affective Influences on Consumer 



Satisfaction with Products," Journal of Consumer Research 7 
(June), 49-54. 

(1980b), "A Rating Scale for Measuring Product/ Service 



Satisfaction," Journal of Marketing 44 (Fall), 68-72. 



449 



(1980c), "Consumer Satisfaction as a Function of Personal 

Competence/Efficacy," Journal of the Academy of Marketing Science 
8 (Fall), 427-437. 

(1981), "Sources of Consumer Satisfaction With Retail 



Outlets," Journal of Retailing 57 (Fall), 68-85, 
(1983), "Consumer Satisfaction and the Phenomenology of 



Emotions During Automobile Ownership Experiences, in 
International Fare in Consumer Satisfaction and Complaining 
Behavior , eds. Ralph L. Day and H. Keith Hunt, Bloomington, IN: 
Indiana University, 2-9. 

(1984), "On Satisfaction Responses and the Effects of 



Product Labels," in Advances in Consumer Research: Volume 11 , ed. 
Thomas C. Kinnear, Provo, UT: Association for Consumer Research, 
31 1-313. 

(1987), "Product/Consumption-Based Affective Responses and 



Postpurchase Processes," Journal of Marketing Research 24 
(August), 258-270. 

and Joseph A. Cote, Jr. (1980), "An Exploratory Study oi 



Non-Product Related Influences Upon Consumer Satisfaction, in 
Advances in Consumer Research: Volume 7 , ed. Jerry C. Olson, Ann 
Arbor, MI: Association for Consumer Research, 577-581. 

and Richard L. Oliver (1981), "Developing Better Measures 

of Consumer Satisfaction: Some Preliminary Results," in Advances 
in Consumer Research: Volume 8 , ed. Kent B. Monroe, Ann Arbor, MI: 
Association for Consumer Research, 94-99. 

and Richard L. Oliver (1984), "Emotion and Consumer 



Satisfaction in Product Ownership/Usage," unpublished manuscript. 

and Michael D. Reilly (1983), "Value-Percept Disparity: An 

Alternative to the Disconf irmation of Expectations Theory of 
Consumer Satisfaction," in Advances in Consumer Research: Volume 
J_0, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Arbor, MI: 
Association for Consumer Research, 256-261 . 

Wilkes, Robert E. (1978), "Fraudulent Behavior by Consumers," Journal 
of Marketing 42 (October), 67-75. 

Wiikie, William L. and Edgar A. Pessemier (1973), "Issues in 

Marketing's Use of Multi-Attribute Attitude Models," Journal of 
Marketing Research 10 (November), 428-441. 

Wilkinson, J. B. and J. Barry Mason (1979), "Dissatisfaction and 

Complaining Behavior as a Result of Unavailable Advertised Food 
Specials," Journal of Consumer Policy 3 (Spring), 43-55. 



450 



Wilson, David T. , H. Lee Mathews, and James W. Harvey ( 1 975) , "An 
Empirical Test of the Fishbein Behavioral Intention Model," 
Journal of Consumer Research 1 (March), 39-48. 

Wilton, Peter C. and David K. Tse (1983), "A Model of Consumer Response 
to Communication and Product Experiences," in Advertising and 
Consumer Psychology , eds. Larry Percy and Arch G. Woodside, 
Lexington, MA: Lexington Books, 315-332. 

Winkler, Robert L. (1972), An Introduction to Bayesian Inference and 
Decision , New York, NY: Holt, Rinehart and Winston, Inc. 

Winter, Frederick W. (1974), "The Effect of Purchase Characteristics on 
Postdecision Product Reevaluation, " Journal of Marketing Research 
11 (May), 164-171. 

Withey, Stephen B. (1977), "Integrating Some Models About Consumer 

Satisfaction," in Conceptualization and Measurement of Consumer 
Satisfaction and Dissatisfaction , ed. H. Keith Hunt, Cambridge, 
MA: Marketing Science Institute, 120-149. 

(1974), "Values and Social Change," in Subjective Elements 

of Weil-Being , ed . Burkhard Strumpel, Paris: Organisation for 
Economic Cooperation and Development. 

Woodruff, Robert B., Ernest R. Cadotte, and Roger L. Jenkins (1983a), 
"Modeling Consumer Satisfaction Processes Using Experience-Based 
Norms," Journal of Marketing Research 20 (August), 296-304. 

, Ernest R. Cadotte, and Roger L. Jenkins (1983b), 

"Charting a Path for CS/D Research," in International Fare in 
Consumer Satisfaction and Complaining Behavior , eds. Ralph L. Day 
and H. Keith Hunt, Bloomington, IN: Indiana University, 118-123. 

Young, Paul T. (1961), Motivation and Emotion: A Survey of the 

Determinants of Human and Animal Activity, New York, NY: John 
Wiley & Sons, Inc. 

Zaichkowsky, Judith Lynne (1985), "Measuring the Involvement 

Construct," Journal of Consumer Research 12 (December), 341-352. 

(1987), "The Emotional Aspect of Product Involvement," 
Advances in Consumer Research: Volume 14 , ed. Melanie Wallendorf 
and Paul Anderson, Provo, UT: Association for Consumer Research, 
32-35. 

Zajonc, Robert B. (1984), "On the Primacy of Affect," American 
Psychologist , 39 (Number 2), 117-123. 

(1980), "Feeling and Thinking: Preferences Need No 
Inferences," American Psychologist 35 (February), 151-175. 



451 



(1968), "Attitudinal Effects of Mere Exposure," reprinted 

in Readings in Social Psychology , eds . Jonathan L. Freedman, J. 
Merrill Carlsmith and David 0. Sears, Englewood Cliffs, NJ: 
Prentice-Hall, Inc., 1971, 93-131. 

Zaltman, Gerald, Rajendra K. Srivastava, and Rohit Deshpande (1978), 
"Perceptions of Unfair Marketing Practices: Consumerism 
Implications," Advances in Consumer Research: Volume 5 , ed. H. 
Keith Hunt, Ann Arbor, MI: Association for Consumer Research, 
247-253. 

Zeithaml, Valarie A. (1982), "Consumer Response to In-Store Price 

Information Environments," Journal of Consumer Research 8 (March), 
357-369. 

(1981), "How Consumer Evaluation Processes Differ Between 

Goods and Services," in Marketing of Services , eds. James Donnelly 
and William George, Chicago, IL: American Marketing Association, 
186-190. 

Zillman, Dolf (1978), "Attribution and Misattribution of Excitatory 
Reactions," in New Directions in Attribution Research, Volume 2 , 
eds. John H. Harvey, William Ickes and Robert F. Kidd, Hillsdale, 
NJ: Lawrence Erlbaum Associates, 335-368. 

Zimbardo, Philip G. (1960), "Involvement and Communication Discrepancy 
as Determinants of Opinion Conformity," Journal of Abnormal and 
Social Psychology 60 (Number 1), 86-94. 

Zimmer, Mary and Rohit Deshpande (1984), "The Development of Choice 
Tactics in Low Involvement Situations," in Proceedings of 1984 
Winter Educators' Conference: Scientific Method in Marketing , 
eds. Paul F. Anderson and Michael J. Ryan, Chicago, IL: American 
Marketing Association, 220-223. 



BIOGRAPHICAL SKETCH 

Douglas R. Hausknecht was born August 10, 1957, in Baltimore, 
Maryland. He is the son of Norman C. and N. Elaine Hausknecht and the 
brother of Brian, Cheryl, and Pamela. 

His education began in Baltimore parochial schools, St. 
Augustine, and was continued in Jacksonville, Florida, where the family 
moved in 1970. He was graduated from Samuel W. Wolf son Senior High 
School in June, 1975. 

His college education followed immediately at the University of 
Florida and culminated in a Bachelor of Science in Agriculture degree 
(major: food science) in March, 1979. Subsequently, in June of 1981, 
he was graduated Master of Business Administration (major: marketing). 
He began doctoral studies in January, 1982. 

He has published articles in the Journal of Consumer Research and 
Ad vances in Consumer Research . He began teaching at the University of 
Akron in January, 1986. 



452 



I certify that I have read this study and that in my opinion, it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 




Richard J. Lutz, Chairman 
Professor of Marketing 



I certify that I have read this study and that in my opinion, it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

Us* /L- 




J/fse^ph W. Alba 
-Associate Professor of 
Marketing 



I certify that I have read this study and that in my opinion, it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 



-/^a/C: 



Esam Ahmed 

Professor of Food Science 

and Human Nutrition 



This dissertation was submitted to the Graduate Faculty of the 
Department of Marketing in the College of Business Administration and 
to the Graduate School and was accepted as partial fulfillment of the 
requirements for the degree of Doctor of Philosophy. 

December 1988 



Dean, Graduate School 



UNIVERSITY OF FLORIDA 



3 1262 08285 323 4